You are on page 1of 96

UMAP

The

Journal

Publisher
COMAP, Inc.

Vol. 24, No. 4

Executive Publisher
Solomon A. Garfunkel
ILAP Editor
Chris Arney
Interim Vice-President
for Academic Affairs
The College of Saint Rose
432 Western Avenue
Albany, NY 12203
arneyc@mail.strose.edu

Editor

On Jargon Editor
Yves Nievergelt
Department of Mathematics
Eastern Washington University
Cheney, WA 99004
ynievergelt@ewu.edu

Paul J. Campbell
Campus Box 194
Beloit College
700 College St.
Beloit, WI 535115595
campbell@beloit.edu

Reviews Editor
James M. Cargal
Mathematics Dept.
Troy State Univ. Montgomery
231 Montgomery St.
Montgomery, AL 36104
jmcargal@sprintmail.com
Chief Operating Ofcer
Laurie W. Aragon

Production Manager
George W. Ward
Director of Educ. Technology
Roland Cheyney
Production Editor
Pauline Wright
Copy Editor
Timothy McLean
Distribution
Kevin Darcy
John Tomicek
Graphic Designer
Daiva Kiliulis

Associate Editors
Don Adolphson
David C. Chris Arney
Ron Barnes
Arthur Benjamin
James M. Cargal
Murray K. Clayton
Courtney S. Coleman
Linda L. Deneen
James P. Fink
Solomon A. Garfunkel
William B. Gearhart
William C. Giauque
Richard Haberman
Charles E. Lienert
Walter Meyer
Yves Nievergelt
John S. Robertson
Garry H. Rodrigue
Ned W. Schillow
Philip D. Strafn
J.T. Sutcliffe
Donna M. Szott
Gerald D. Taylor
Maynard Thompson
Ken Travers
Robert E.D. Gene Woolsey

Brigham Young University


The College of St. Rose
University of HoustonDowntown
Harvey Mudd College
Troy State University Montgomery
University of WisconsinMadison
Harvey Mudd College
University of Minnesota, Duluth
Gettysburg College
COMAP, Inc.
California State University, Fullerton
Brigham Young University
Southern Methodist University
Metropolitan State College
Adelphi University
Eastern Washington University
Georgia College and State University
Lawrence Livermore Laboratory
Lehigh Carbon Community College
Beloit College
St. Marks School, Dallas
Comm. College of Allegheny County
Colorado State University
Indiana University
University of Illinois
Colorado School of Mines

Subscription Rates for 2004 Calendar Year: Volume 25


Membership Plus
Individuals subscribe to The UMAP Journal through COMAPs Membership Plus. This subscription
also includes print copies of our annual collection UMAP Modules: Tools for Teaching, our
organizational newsletter Consortium, on-line membership that allows members to download and
reproduce COMAP materials, and a 10% discount on all COMAP purchases.

(Domestic)
(Outside U.S.)

#2420
#2421

$90
$105

Institutional Plus Membership


Institutions can subscribe to the Journal through either Institutional Plus Membership, Regular
Institutional Membership, or a Library Subscription. Institutional Plus Members receive two print
copies of each of the quarterly issues of The UMAP Journal, our annual collection UMAP Modules:
Tools for Teaching, our organizational newsletter Consortium, on-line membership that allows
members to download and reproduce COMAP materials, and a 10% discount on all COMAP purchases.

(Domestic)
(Outside U.S.)

#2470
#2471

$415
$435

Institutional Membership
Regular Institutional members receive print copies of The UMAP Journal, our annual collection
UMAP Modules: Tools for Teaching, our organizational newsletter Consortium, and a 10% discount on
all COMAP purchases.

(Domestic)
(Outside U.S.)

#2440
#2441

$180
$200

Web Membership
Web membership does not provide print materials. Web members can download and reproduce
COMAP materials, and receive a 10% discount on all COMAP purchases.

(Domestic)
(Outside U.S.)

#2410
#2410

$39
$39

To order, send a check or money order to COMAP, or call toll-free


1-800-77-COMAP (1-800-772-6627).

The UMAP Journal is published quarterly by the Consortium for Mathematics and Its Applications
(COMAP), Inc., Suite 210, 57 Bedford Street, Lexington, MA, 02420, in cooperation with the
American Mathematical Association of Two-Year Colleges (AMATYC), the Mathematical
Association of America (MAA), the National Council of Teachers of Mathematics (NCTM), the
American Statistical Association (ASA), the Society for Industrial and Applied Mathematics (SIAM),
and The Institute for Operations Research and the Management Sciences (INFORMS). The Journal
acquaints readers with a wide variety of professional applications of the mathematical sciences and
provides a forum for the discussion of new directions in mathematical education (ISSN 0197-3622).
Periodical rate postage paid at Boston, MA and at additional mailing offices.

Send address changes to: info@comap.com


COMAP, Inc. 57 Bedford Street, Suite 210, Lexington, MA 02420
Copyright 2003 by COMAP, Inc. All rights reserved.

Vol. 24, No. 4

2003

Table of Contents
Editorial
We Have Already Been to the Future
Paul J. Campbell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393

Articles
Efficient Signal Transmission and Wavelet-Based Compression
Susan E. Kelly and James S. Walker . . . . . . . . . . . . . . . . . . . . . . . .395

UMAP Modules
The Spread of Forest Fires
Emily E. Puckette . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 417

ILAP Modules
Getting the Salt Out
Ethan Berkove, Thomas Hill, Scott Moor . . . . . . . . . . . . . . . . . . . 435
Vehicle Emissions
Lei Yu and Don Small. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .451

Reviews . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .473
Annual Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 479
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 483
Errata . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 484

Editorial

393

Editorial
We Have Already Been to the Future
Paul J. Campbell
Mathematics and Computer Science
Beloit College
Beloit, WI 53511
campbell@beloit.edu
Every 10 years or so, the Mathematical Association of America publishes
recommendations for the undergraduate program in mathematics; long ago,
I was part of a panel that prepared such a report. The last report (1991) was
short; the mathematics community then was in the throes of the calculus wars,
computer science was coming on strong, and there was no longer common
agreement on what courses should be part of a mathematics major.
After four years of meetings, panel discussions, focus groups, conferences,
surveys of mathematics departments, consultation with partner disciplines,
and deliberations, there is a new curriculum guide [CUPM 2003].
The new report distinguishes itself by not focussing primarily on the program for mathematics majors but addressing the entire college-level mathematics curriculum, for all students, even those who take just one course.
The six recommendations of the report are not likely to be controversial,
and they are not accompanied by implementation suggestions. Here are paraphrases:
1. Understand and meet the needs of the students, and assess the effectiveness
of doing so.
2. Every course should have activities that promote analytical reasoning, problemsolving, communication skills, and mathematical habits of mind.
3. Every course should present various perspectives, employ examples and
applications from contemporary topics, and promote students ability to
apply the material.
4. Mathematical sciences departments should collaborate with other departments on courses, majors, research projects, and teaching.
c
The UMAP Journal 24 (4) (2003) 393394. Copyright
2003 by COMAP, Inc. All rights reserved.
Permission to make digital or hard copies of part or all of this work for personal or classroom use
is granted without fee provided that copies are not made or distributed for prot or commercial
advantage and that copies bear this notice. Abstracting with credit is permitted, but copyrights
for components of this work owned by others than COMAP must be honored. To copy otherwise,
to republish, to post on servers, or to redistribute to lists requires prior permission from COMAP.

394

The UMAP Journal 24.4 (2003)

5. At every level of the curriculum, some courses should have activities that
teach the use of technology in solving problems and understanding mathematical ideas.
6. Faculty teaching and curricular efforts should be rewarded.
The report cites
changes:
declining numbers of bachelors degrees in mathematics (down 20%
since 1990 while total bachelors degrees rose 20%),
corresponding drops in enrollments in calculus and in advanced courses,
and
a high proportion (30%) of mathematics majors having a second or joint
major (i.e., they are interested primarily in something other than mathematics);
a nagging constant: Mathematics has extremely poor public relations. Students, as well as deans and faculty from other departments,view
mathematicians as insular and excessively inward looking;
mathematics departments as disconnected from other disciplines except
through a service component which they believe is accepted only reluctantly and executed without inspiration or effectiveness, and uninterested in adapting instruction to needs of non-majors; and
(crucially) mathematics itself as irrelevant for them (as in math is tough
and too much work anyway, so Ill do without it but keep my self-esteem
by deeming it irrelevant).
COMAP has been ahead of the curveor at least, the recommendations
are catching up with what COMAP has been doing all along. For more than
25 years, COMAP has played a major role in developing and disseminating
materials that promote the use of contemporary applications in teaching mathematics. ILAPs (see the two in this issue) in particular encourage collaboration
with faculty in other disciplines.
COMAP alone cannot change mathematical instruction or societal attitudes.
However, readers of this Journal need to direct the attention of their colleagues
(particularly those new to the profession) to the resources that COMAP offers
for implementing the new curricular recommendations.

References
Committee on the Undergraduate Program in Mathematics (CUPM). 2003. Undergraduate Programs and Courses in the Mathematical Sciences: CUPM Curriculum Guide 2004; http://www.maa.org/cupm/welcome.html . Washington,
DC: Mathematical Association of America.

Efficient Signal Transmission

395

Efcient Signal Transmission


and Wavelet-Based Compression
Susan E. Kelly
Department of Mathematics
University of WisconsinLa Crosse
La Crosse, WI 54601
kelly.susa@uwlax.edu

James S. Walker
Department of Mathematics
University of WisconsinEau Claire
Eau Claire, WI 54702
walkerjs@uwec.edu
Abstract
We describe how digital signals can be transmitted by analog carrier signals. This is
the method of pulse code modulation. Transmission by pulse code modulation is widely
used thanks to its abilities to withstand corruption by noise and to transmit multiple signals simultaneously. Digital signals can also be used to compress analog signals, thus
allowing for more rapid transmission, by employing wavelet series. We shall describe
how wavelet series are employed and compare them with Fourier series, showing that in
many instances wavelet series provide a better technique.

Introduction
A common method of transmitting digital signals is pulse code modulation (PCM).
A PCM signal is an analog signal that encodes the bits of the digital signal via a carrier
wave. The frequency of this carrier wave is modified in such a way that the bits can be
read off at the receiving end. This PCM method has a couple of beautiful features.
It can withstand large amounts of noise in the environment that typically corrupt
the signal by the time it reaches the receiver.
It allows for simultaneous transmission of several different messages using just one
PCM signal.
c
The UMAP Journal 24 (4) (2003) 395416. Copyright
2003 by COMAP, Inc. All rights reserved.
Permission to make digital or hard copies of part or all of this work for personal or classroom use is granted
without fee provided that copies are not made or distributed for profit or commercial advantage and that
copies bear this notice. Abstracting with credit is permitted, but copyrights for components of this work
owned by others than COMAP must be honored. To copy otherwise, to republish, to post on servers, or to
redistribute to lists requires prior permission from COMAP.

396

The UMAP Journal

24.4 (2003)

An important benefit of digital signals is their ability to compress analog signals. Instead of transmitting the entire analog signal, it is more efficient to transmit numbersdigitally encodedthat describe how to reconstruct the signal at the
receiving end. A classical method for producing these numbers is to compute coefficients in a Fourier series. Recently, a whole new approach, called wavelet series,
has been developed. Coefficients of a wavelet series are computed, digitally encoded,
then transmitted in digital form. At the receiving end, these coefficients are used to
produce a wavelet series that provides an excellent approximation to the original signal. Wavelet series are often more efficientfewer coefficients are needed for a good
approximationthan Fourier series.
In the next section we shall describe the theory behind PCM signals and illustrate
their application to digital signal transmission. We then discuss Fourier series and
wavelet series and their application to compression of analog signals. The paper concludes with a brief survey of references where further details and extensions of these
ideas can be found.
The numerical data and graphs in this paper were produced with the free software
FAWAV [Walker 2000; 2002].

PCM and Digital Signal Transmission


In this section we shall explain how an analog carrier signal can be modified to encode a sequence of bits. Suppose, for example, that we want to transmit the following
sequence of seven bits:
1 0 0 0 1 1 0.

(1)

In order to represent any of the bits having value 1, we use a signal of the form:
2

g(t) = e(tk) cos 2t.

(2)

Figure 1a shows a graph of this signal for k = 6, = 2, and = 6; Figure 1b shows


a different graph, using k = 3, = 25, and = 12. It is not hard to see that the

Figure 1. a. Pulse centered at position t = 6. b. Pulse centered at position t = 3.


2

parameter k controls where the center of the amplitude factor e(tk) lies. Also, the
parameter controls how quickly this factor damps down toward 0 as t moves away

Efficient Signal Transmission

397

from k; in Figure 1b, is greater than in Figure 1a, hence the amplitude of the pulse
in Figure 1b damps down more quickly to 0.
The parameter in (2) determines the frequency of the factor cos 2t. Hence, the
cosine factors for the pulses shown in Figure 1 have frequencies of = 6 and = 12.
As you can see from the figure, the oscillations of the pulse in Figure 1b are faster
(twice faster, in fact) than the oscillations of the pulse in Figure 1a.
It is a standard task in electrical engineering to design a circuit having maximum
response centered on a desired frequency [Boylstad 1994; Floyd 2001]. For example,
if we want to send the bit 1 using the PCM signal shown in Figure 1b, then we should
design a receiver having an electric circuit element with maximum response at = 12
(Figure 2a). The response curve in Figure 2b would provide an excellent detector
for such a transmission of the bit 1. Likewise, if we wanted to use the PCM signal in
Figure 1a, then the maximum response for the receiver should be centered at = 6.

Figure 2. Response curves for electric circuits with maximum response at = 12. a. A broad response
curve, unsuitable for precise detection. b. A narrow, precisely tuned response curve, suitable for accurate
detection.

By combining both such circuit elements, we could even have a receiver that could
detect both bits, transmitted simultaneously. We shall return to this point later.
Now that we have discussed how to transmit a single bit having value 1, we can
turn to the question of how to transmit a sequence of bits having different values, as
in (1).

Transmitting a Bit Sequence


The method of transmitting a bit sequence, such as the one shown in (1), is to
simply add together signals of the type shown in (2). One frequency, 1 , is used for
transmitting bits of value 1, and a second frequency 0 = 1 is used for transmitting
bits of value 0.
For example, to transmit the sequence of bits 1 0 0 0 1 1 0, we use the following
PCM signal:


(t2)2
(t3)2
(t4)2
(t7)2
+e
+e
+e
f (t) = e
cos 20 t


2
2
2
+ e(t1) + e(t5) + e(t6) cos 21 t.

398

The UMAP Journal

24.4 (2003)

with = 150, 0 = 232, and 1 = 280 (Figure 3a). In Figures 3 and 4, we illustrate
the scheme for decoding at the receiving end the seven bits from the PCM signal.

Figure 3. Filtering of PCM signal to process individual pulses. The filters w1 and w4 in b are used to cut
out the individual pulses at positions t = 1 and t = 4. The cutout pulses are shown in c and d.

To isolate an individual pulse at position t = k, the PCM signal f (t) is multiplied


by a filter (or window) function wk (t). This function wk (t) is equal to 1 in an interval
(k , k + ) centered at k, and decreases smoothly down to 0 for t < k 1 + and
t > k + 1 . See Figure 3b, where the filters w1 and w4 are shown. Multiplying
the PCM signal f by the filter wk produces a single pulse f (t)wk (t) that is nonzero
only within the interval [k 1 + , k + 1 ] centered at t = k. In Figures 3c and
3d, we show the pulses produced by multiplying the PCM signal f by w1 and w4 . The
multiplications produce single pulses centered at t = 1 and t = 4.
In Figure 4b, we show the whole set of filters {wk }8k=0 for detection of the pulses
at positions t = 0, 1, 2, . . . , 8. These filters satisfy
8


wk (t) = 1

(3)

k=0

for 0 t 8. Equation (3) implies that the PCM signal f (t) is equal to the sum of the
filter functions f (t)wk (t) over the interval [0, 8]. For decoding longer PCM signals,
(3) needs to be extended via a larger set of filters, say {wk }N
k=0 , satisfying
N

k=0

for some large positive integer N .

wk (t) = 1

(4)

Efficient Signal Transmission

399

Figure 4. PCM transmission of bit sequence 1 0 0 0 1 1 0. a. PCM signal. b. Filter bank used for detection.
c. Detection: [bottom] PCM signal, [top] magnitude of responses to filter bank output. Darker pixels
represent more intense response.

While (4) is a handy property, it is not absolutely necessary for the decoding
scheme to work. In fact, for decoding to work well it is sufficient that the following double inequality hold:
A

N


wk (t) B,

k=0

where A and B are positive constants. Often a value of A that is significantly larger
than 1 is used, producing an amplification of the PCM signal. Amplification is often
needed because PCM signals are significantly reduced in amplitude after transmission
over long distances.
In Figures 3c and 3d, the individual pulses are shown as having restricted domains:
[, 2 ] for w1 and [3 + , 5 ] for w4 . The domain of each pulse f (t)wk (t)
is restricted to facilitate processing by the frequency detector. Since f (t)wk (t) = 0
outside of the interval [k 1 + , k + 1 ], there is no need for the detector to process
this pulse outside of this interval.
In Figure 4c, we show the result of processing each isolated pulse by a frequency
detector. The PCM signal is shown at the bottom of the figure. The top of the figure

400

The UMAP Journal

24.4 (2003)

shows the magnitude of the response of a frequency detectora frequency detector


that sweeps through a set of tuning frequencies {2, 4, 6, . . . , 1024}, just like a dial on a
radio. This set includes the frequencies 0 = 232 and 1 = 280 used for transmitting
bits of value 0 and 1, respectively. (The reason for displaying responses for a large
range of frequencies will be clearer after we discuss simultaneous transmission.) As
the reader can see, the bit sequence 1 0 0 0 1 1 0 is easily read off from the display of
the detector response.

Simultaneous Messages and Denoising


The beauty of PCM signal transmission consists in its wealth of capabilities. In
particular, it provides a simple way of transmitting several messages simultaneously
using just one PCM signal. Furthermore, it allows for accurate recovering of signals
even after the addition of large amounts of noise. Sometimes the noise is so severe that
it will not be apparent from the graph of the signal that any transmission has occurred
at all; nothing but noise will be visible. Yet, upon decoding, the transmitted bits will
stand out clearly from the noise. We now provide examples illustrating these ideas.

Figure 5. Decoding three simultaneous messages.

Example: We show how the three bit sequences 1 0 0 1 1 0 0, 1 1 0 0 1 1 1, and


0 1 0 1 0 0 1 can be transmitted simultaneously. The method consists of adding
together three PCM signals using three different values of 0 and 1 : We use
0 = 768 32, 1 = 768 + 32; 0 = 512 32, 1 = 512 + 32; and 0 = 256
32, 1 = 256 + 32, respectively. The resulting signal is shown at the bottom,
and its decoding at the top, of Figure 5. The three separate messages have been
decoded from a single PCM signal. With present technology, several thousand
separate messages can be transmitted with just one PCM signal [Catala 1997].

Efficient Signal Transmission

401

Example: We simulate the recovery of a bit sequence from a PCM signal that
has been severely corrupted by additive noise. In Figure 6a, we show an example of a PCM signal for transmitting the bit sequence 1 1 0 0 1 0 1. The frequencies 0 = 480 and 1 = 544 were used for transmitting the bits. We added
to this signal a simulation of Gaussian (normally distributed) random noise with
standard deviation 1. Gaussian noise is frequently used for modeling real-world
noise because the Central Limit Theorem implies that sums of noise from many
independent sources, over the course of transmission, yield a total noise that is
approximately Gaussian. A discrete realization of Gaussian noise with a standard deviation 1 is shown in Figure 6b, and a histogram of its values is shown
in Figure 6c. This histogram illustrates the Gaussian distribution underlying
the random noise. The addition of the noise to the PCM signal is shown in Figure 6d. Notice that it is impossible to tell from this noisy signal that any message
has been transmitted at all.

Figure 6. Simulation of additive noise corrupting a PCM signal.

The decoding of the noisy PCM signal is shown in Figure 7. In Figure 7a,
we can see the responses to the bits in the PCM signal as prominent black
smudges, which stand out from a light grey background resulting from the noise.
These smudges are fairly easy for our eyes to decode as the bits 1 1 0 0 1 0 1
encoded by the original PCM signal. For electronic recognition, however, it is
better if frequency responses having magnitude below some threshold are set to
zero. In Figure 7b, we show such a thresholded response. The responses to the
bits are easily detectable electronically.

402

The UMAP Journal

24.4 (2003)

Figure 7. a. Decoding noisy message. b. Thresholded decoding.

Compressing Signals
We have seen how to transmit a signal if we can represent it as a sequence of
bits. In this section, we consider three different ways to obtain such a sequence, using
power series, Fourier series, and wavelet series. We shall see why wavelet series often
provide the best method.

Power Series
In calculus, we see how a function can be represented by a power series. The
Taylor series of a function f (t) at t = a is given by


f (n) (a)
(t a)n .
n!
n=0
n
M
If limM |f (t) n=0 f n!(a) (t a)n | = 0 for |t a| < R, we say that f is equal
to the sum of its Taylor series on the interval described by |t a| < R. For example,
the Taylor series for f (t) = sin t + 2 cos 2t centered about the origin is

sin t + 2 cos 2t =





cn t2n+1 + dn t2n ,

n=0

where
cn =

(1)n
(2n + 1)!

and

dn =

(1)n 22n+1
(2n)!

Efficient Signal Transmission

403

for all real t. Thus, to transmit the function f (t), one need know only the coefficients
cn and dn of this power series. These coefficients can be written as 16-bit binary codes.
These codes can then be strung together to give bit sequences.

Figure 8. f (t) = sin t + 2 cos 2t.

Of course, in practice only a finite number of coefficients can be used; and when
we truncate the series by using only the first N terms, we introduce error in our representation. For each t-value, this error would be



  f (n) (a)

n
(t

a)
.



n!
n=N

Figure 8 shows the graph of the function f (t) = sin t + 2 cos 2t. The Taylor series for
this function at t = 0 converges for all values of t; but when we only use a finite number of terms in the series, we obtain errors. The series works best near t = 0, the point
about which the series is being expanded. As we add more terms, the reconstructed
series resembles the function for a larger domain; but since we are adding terms of
the form tn , at some point the reconstructed function will increase or decrease without
bound. These ideas are illustrated in Figure 9.

Figure 9. Taylor series at t = 0 for sin(t) + 2 cos(2t). Left: 10 terms. Right: 50 terms. The original
function is shown in a lighter line.

For a Taylor series about y = a, we can represent a function in terms of powers of


t a. Still, even when the power series converges everywhere, a finite sum may not
represent the function well for values a distance away from t = a (Figure 9).

404

The UMAP Journal

24.4 (2003)

Another problem with Taylor series in many applications is that computing the
derivatives needed for the coefficients is not always practical or possible. What we
need in practical applications is a series with more easily computed coefficients and
which requires fewer terms to give a good reconstruction of our original signal. We
shall see how Fourier series improve our work in both of these respects, and how to
compress data still further using wavelets.

Fourier Series
Another way to represent many functions was introduced by Jean B.J. Fourier in
the early nineteenth century. A Fourier series represents a periodic function in terms
of sine and cosine functions. For a 2W -periodic function f (t), its Fourier series is the
series in the following correspondence:






nt
nt
1
an cos
+ bn sin
,
f (t) a0 +
2
W
W
n=1
where the Fourier coefficients are defined to be


W
1
nt
an =
dt,
f (t) cos
W W
W

W
1
nt
bn =
f (t) sin
dt,
W W
W

for n = 0, 1, 2, ...;
for n = 1, 2, ....

Note that the 2-periodic signal, sin t + 2 cos 2t, from the previous section can be
captured exactly with only two nonzero Fourier coefficients!
A Fourier series can also be written in the complex-number form

f (t)

cn eint/W ,

n=

where the Fourier coefficients are


1
cn =
2W

f (t)eint/W dt.

One can view a Fourier series as a sum of terms having single frequencies, the amplitudes of those terms are the Fourier coefficients for the function. Fourier series are
expansions of functions using the basis functions



 nt 
nt
nt
, cos
1, sin
or
ei W
,
W
W
nZ
n=1
which are orthogonal bases of periodic functions on [W, W ].

Efficient Signal Transmission

405

Most signals are not periodic, however, and so we need to broaden


 the notion of
1
Fourier series. For piecewise smooth functions in L (R) = {f : |f (t)| dt < },
this can be done via the Fourier transform

f() =
f (t)ei2t dt
(5)

and the Fourier inversion formula



f (t) =

f()ei2t d.

The discrete form of Fourier series for analog signals is a consequence of the Sampling Theorem.
Theorem 1 (Sampling Theorem) If a continuous function f (t) L1 (R) is bandlimited, that is, f() is nonzero only on some interval [W, W ], then f (t) is completely
determined by a sequence of points spaced 1/2W units apart. Specifically,

f (t) =

n=

where sinc(t) =

 n 
sinc(2W t n),
2W

sin(t)
.
t

Discussions of Theorem 1 can be found in Higgins [1985], Kelly [2000], and Walker
[1988].
n
)}, to create
Theorem 1 allows us to use sample values of our function, {f ( 2W
a bit sequence. However, we are also interested in compressing our data by creating
as short a bit sequence as possible while still retaining the ability to generate a good
approximation to the original function. To explore this issue further, we investigate the
discrete Fourier transform.
In practice, our function f (t) will exist only for a finite duration of time. Therefore,
we can scale the function so that f (t) = 0 for t  [0, 1]. We can then apply Theorem 1
to f() in place of f (t). Hence, f() can be recovered from the set {f(k)}kZ .
Using the definition of the Fourier transform in (5) and the fact that f (t) is zero
away from [0, 1], we obtain
f(k) =

f (t)ei2kt dt

= lim

N
1 

n=0

n
N

ei2k N

 1
.
N

This last sum comes from the definition of a Riemann integral. Thus, for N large,
f(k) Fk ,

(6)

406

The UMAP Journal

24.4 (2003)

where
Fk =

N 1
1   n  i2nk/N
f
e
N n=0
N

for all k Z


 n N 1
is the N -point discrete Fourier transform (DFT) of the sequence fn = f N
.
n=0
Formula (6) shows how the DFT values are determined by our functions sample values. Now let us investigate how to recover our sample values, and hence our
function, from a finite number of DFT values. We begin with the finite sum

N
1
N
1
N
1



j
1
ei2jk/N ei2nk/N
Fk ei2nk/N =
f
N j=0
N
k=0
k=0



N 1 N 1
1   i2/N (jn)k
j
,
=
(e
)
f
N j=0
N
k=0

for n = 0, 1, 2, ..., N 1. Note that (ei2/N )(jn) is a nontrivial root of


xN 1 = (x 1)(xN 1 + xN 2 + ... + x + 1)
N 1
for j n = 0, and thus k=0 [(ei2/N )(jn) ]k = 0 for j = n. From this, we obtain
N
1

k=0

Fk ei2nk/N = f

n
N

Thus to obtain all needed information about our function, we need to store either the
1
N 1
n
)}N
N samples {fn = f ( N
n=0 or the N DFT values {Fk }k=0 .
Because of the desire to compress our data, we hope to use only a fraction of the
DFT valuesthose with largest magnitudesand still get a good reconstructed function. Unfortunately, the magnitudes of the DFT values often damp down very slowly;
hence, we often need most, if not all, of these coefficients to obtain a good reconstruction of our original signal. We provide examples in the next subsection, when we
compare this DFT-based method of compression with wavelet-based compression.

Wavelets
In an age when we seek faster and sharper scanners for medical diagnosis, moreefficient computer storage techniques for larger amounts of computer data, and higherfidelity audio and visual signals with less information used, a more efficient way to
convert analog signals to digital ones is needed. Many people working in harmonic
analysis helped to provide the groundwork for the wavelet bases, which were introduced in the 1980s. Wavelets are beneficial in the areas just mentioned and in many
other applications because of the highly efficient ways a wavelet series can represent a
function. As we illustrate below, they often allow for a better-reproduced analog signal
using less data than with Fourier techniques.

Efficient Signal Transmission

407

Let us begin by defining a wavelet


 basis.2 We will work with wavelet bases2for the
2
set of functions L (R) = {f : |f (t)| dt < }. A wavelet basis of L (R) is
a basis of functions generated using dilations by 2m and translations by n2m of a
single function. Here m and n are both integers. That is, if (t) is such a generating
function, then


m,n (t) = 2m/2 (2m t n)
m,nZ

is a wavelet basis of L2 (R). By a basis, we mean that for each f (t) L2 (R), the
following holds:
 
nm m,n (t)
f (t) =
mZ nZ

where
nm

f (t)m,n (t) dt

are the wavelet coefficients. We shall refer to as the generating wavelet for this
wavelet basis.
A related way to represent our function f (t) is with scaling functions generated
by a single function . For any integer M , the set {m,n }m<M,nZ is a basis of
a subset VM L2 (R). We can use (t) to generate another basis for VM , the set
{M,n (t) = 2M/2 (2M t n)}nZ . Since limM VM = L2 (R), it follows that for
M sufficiently large

nM M,n (t),
f (t)
nZ

where
nM

f (t)M,n (t)dt.

(7)

We refer to as the generating scaling function for these bases {M,n (t)}nZ . We
see below how both bases are used when we examine the fast wavelet transform.
There are an infinite number of wavelet bases for L2 (R). The Haar wavelet system is undoubtedly the simplest one. For the Haar system, (t) is the characteristic function [0,1) (t). The generator for the Haar wavelets is defined by (t) =
[0,1/2) (t) [1/2,1) (t). The space VM is the set of all L2 (R) functions which are
constant on each interval [k/2M , (k + 1)/2M ), for all integers k.
The generators and for the Haar system satisfy the following identities:
(t) = (2t) + (2t 1),
(t) = (2t) (2t 1).

408

The UMAP Journal

24.4 (2003)

Substituting 2m1 t into these identities and multiplying by 2

m1
2

yields




m
1  m
2m1 t = 2 2 (2m t) + 2 2 (2m t 1)
2
 m1 

m1
m
1  m
t = 2 2 (2m t) 2 2 (2m t 1) .
2 2 2
2
2

m1
2

From these last two identities, we obtain the following relations between Haar scaling
coefficients and wavelet coefficients:
1 m
+
km1 = 2k
2
1 m

km1 = 2k
2

1 m
2k+1
,
2
1 m
2k+1
.
2

(8)

These identities show that the (m 1)st level coefficients {km1 }, {km1 } are obtained from the mth level scaling coefficients {km }.
If we assume that N is a power of 2, say N = 2R , then the fact that is supported
on [0, 1) implies that when M = R the support of each function M,n (t) will only
cover one sample point n/N . Hence, in view of (7) we are justified in assuming that
nR = 2R/2 fn . Thus, the scaling coefficients at the Rth level can be set equal to a
constant times fn . In practice, the constant factor 2R/2 is dropped and we assume
that nR = fn for each n. The reason for dropping the constant factor 2R/2 is that
then (8), for m = R, defines an orthogonal matrix transformation on each pair of
adjacent sample values f2k and f2k+1 .
We have now shown that, given the scaling coefficients {nR = fn } at the Rth
level, we can use (8) repeatedly to derive all scaling coefficients {nm } and wavelet
coefficients {nm } for all levels m < R. Given the initial data

 R R
R
0 , 1 , . . . , N
1 = (f0 , f1 , . . . , fN 1 ),
we apply (8) to get


R1
R1
0R1 , . . . , N/21
, 0R1 , . . . , N/21
.

For example, if (f0 , f1 , . . . , f7 ) = (2, 2, 2, 2, 4, 4, 4, 12), then R = 3, and using (8)


yields

 2 2 2 2 2 2 2 2 
0 , 1 , 2 , 3 , 0 , 1 , 2 , 3 = 2 2, 2 2, 4 2, 8 2, 0, 0, 0, 4 2 .


R1
We may then repeat application of (8) on the scaling coefficients 0R1 , . . . , N/21
from these last data to get


R2
R2
R1
, 0R2 , . . . , N/41
, 0R1 , . . . , N/21
.
0R2 , . . . , N/41
In the example above, this second application of (8) yields

 1 1 1 1 2 2 2 2 
0 , 1 , 0 , 2 , 0 , 1 , 2 , 3 = 4, 12, 0, 4, 0, 0, 0, 4 2 .

Efficient Signal Transmission

409

Continuing in this way, we obtain at the Rth step the data




R1
.
00 , 00 , 01 , 11 , . . . , N/21
For the example, we can perform one more application of (8), thereby obtaining


 0 0 1 1 2 2 2 2 
0 , 0 , 0 , 2 , 0 , 1 , 2 , 3 = 8 2, 4 2, 0, 4, 0, 0, 0, 4 2 .
m
Since (8) involves only six arithmetic operations on two adjacent coefficients 2k
m
and 2k+1
, it follows that there is a very fast computer algorithm, a fast wavelet transform WH , that performs the transformation:


WH
R1
(f0 , f1 , . . . , fN 1 )
00 , 00 , 01 , 11 , . . . , N/21
.
(9)

Note also that (8) is invertible:


1
m
2k
= km1 +
2
1
m
= km1
2k+1
2

1
km1 ,
2
1
km1 .
2

Hence, the fast wavelet transform in (9) is invertible, by another very fast computer
1
:
algorithm, which we denote by WH


 W 1
H
R1

(f0 , f1 , . . . , fN 1 ).
00 , 00 , 01 , 11 , . . . , N/21


Since (t) dt = 0 and is supported on the interval [0, 1), it follows that
when f (t) is constant on the interval [2m n, 2m (n + 1)), then nm will be zero. The
transformation



WH
8 2, 4 2, 0, 4, 0, 0, 0, 4 2
(2, 2, 2, 2, 4, 4, 4, 12)
illustrates this phenomenon in an elementary way. A more complex illustration is
shown in Figure 10, where in a we show 1,024 samples of the signal
f (t) =
5(sin 8t)(cos 4t) /5,

0 t < 1,

where
denotes the greatest integer function. The signal f (t) is a step function,
constant over several subintervals of [0, 1). The fast Haar transform of the sample data
(fn ) will produce a large number of zeros. Consequently, a good approximation of the
original signal can be obtained with just a fraction of the nonzero Haar coefficients.
1
Figure 10b shows the graph of WH
applied to the 195 highest magnitude Haar coefficients (all other coefficients set equal to zero). This reconstructed signal is a good
approximation of the original.

410

The UMAP Journal

24.4 (2003)

One way to quantify the accuracy of the approximation is to use relative R.M.S. er1
N 1
ror. Given a discrete signal {fj }N
j=0 and its reconstruction {gj }j=0 , the relative
R.M.S. error is defined by

 N 1
 j=0 |fj gj |2
.
D(f, g) =
N 1
2
j=0 |fj |
For convenience, we refer to D(f, g) as rel. error. As a standard of good approximation, we require that a reconstructed signal approximate the original with a rel. error of
at most 102 . The reconstructed Haar signal in Figure 10b approximates the original
signal with the required rel. error of 102 . By comparison, we show a discrete Fourier
approximation in Figure 10c, using the top 195 highest magnitude DFT values. The
rel. error is then 2.6102 . This increase in rel. error is reflected in the ragged appearance of the Fourier approximation. In order to obtain a rel. error of 102 , the Fourier
approximation required 448 coefficients. For this step function signal, the Haar series
clearly outperforms the Fourier method in all respects.

Figure 10. a. Step function data, 1024 points. b. Haar series, 195 highest magnitude coefficients: rel. error
102 . c. Fourier approximation, 195 highest mag. coefficients: rel. error 2.6 102 . d. Coif18 series,
195 highest mag. coefficients: rel. error 3.2 102 .

The simple nature of the functions in the Haar system makes it easy to study the
structure of Haar wavelets. Their lack of smoothness, however, makes them a poor
basis for many applications. While they performed well in the step function example
just considered, this same example illustrates that Haar series produce step function
approximations for all types of signals. If our samples are taken from a smoothly
varying function, then the step function approximations provided by Haar series will

Efficient Signal Transmission

411

be unsuitable. In our work, we shall instead use Coiflets, which are wavelet bases of
I. Daubechies [1992] and R. Coifman (see Burrus et al. [1998]).
The Coiflets are an infinite class of wavelets for L2 (R). Similar to the Haar
wavelets, these wavelets have compact support: for each such system the generating
scaling function (t) and generating wavelet (t) are zero outside of certain intervals
of finite length. Also, just as with the Haar system, we have


(t) dt = 1 and
(t) dt = 0.

These last two integral conditions are true for all wavelets on L2 (R). The added
benefits of Coiflets are their smoothness and the vanishing moment conditions:


k
t (t)dt = 0 and
tk (t)dt = 0

for some positive integers k.

Figure 11. a. Coif18 generating scaling signal. b. Coif18 generating wavelet.

For the remainder of our work, we shall use the Coiflets known as the Coif18
system. For the Coif18 system, both the generating scaling function (t) and the generating wavelet (t) are supported on the interval [6, 11]. Figure 11 shows graphs
of and . These Coif18 generating functions are twice continuously differentiable.
That implies that any partial sum for a Coif18 wavelet series will be twice continuously differentiable as well. Such smoothness for Coif18 partial sums will provide
better approximations for smoothly varying signals than the step function partial sums
provided by Haar series. Besides smoothness, the Coif18 generating wavelet (t) has
six vanishing moments,

tk (t) dt = 0, 0 k 5,

and the Coif18 generating scaling function (t) also has six vanishing moments,

tk (t) dt = 0, 1 k 6.

412

The UMAP Journal

24.4 (2003)

The benefits of vanishing moments


 can be seen when computing the wavelet coefficients. For instance, since (t) dt = 0, if f (t) is constant on the interval
[2m (n 6), 2m (n + 11)), then nm will be zero. Similarly, all of the vanishing
moments for imply that if f (t) is a polynomial of degree less than 6 on this same
interval, then nm will also be zero. Put another way, if f can be closely approximated
by a polynomial of degree less than 6 on an interval containing the support of m,n ,
then nm will be approximately zero. We shall see that these facts are useful for compressing signals. The vanishing of the moments of the generating scaling function is
important for making the approximation nR 2R/2 fn which we discussed above.
This approximation is much better justified for Coif18 scaling functions than for Haar
scaling functions (see Resnikoff and Wells [1998] for more details), and it justifies
replacing (nR ) by the signal samples (fn ).
With this background for the Coif18 system, we can now describe how Coiflets
allow us to compress signals. As with the Haar wavelet system, the Coif18 system
allows for the computation of {nm1 } and {nm1 } by applying an orthogonal transformation to a small number (18, to be precise) of adjacent values of {nm }. (The small
number of adjacent values needed is another advantage of using Coif18 wavelets.) Iterating thisprocess R times
 yields a fast wavelet transform WC , which converts the
R
R
initial data 0 , . . . , N 1 = (f0 , . . . , fN 1 ) as follows:
W

C
R1
(00 , 00 , 01 , 11 , . . . , N/21
).
(f0 , f1 , . . . , fN 1 )

This transformation WC has a fast inverse WC1 , as well (see Walker [1997; 1999]
for more details). Since the Coif18 wavelet has six vanishing moments, many of the
wavelet coefficients nm are often zero or very small. This will give us our desired
compression. We retain only coefficients that have significant size; all other coefficients are set to zero and are not transmitted. These high-magnitude coefficients will
give us our binary codes for transmission. Binary encoding only a small set of highmagnitude coefficients will often produce a set of data much smaller in size than the N
original signal values. At the receiving end, we use only these high-magnitude coefficients, along with zeros inserted for the coefficients we have dropped, and apply WC1
to reconstruct a wavelet series approximation to our signal.
We can illustrate such compression with the alternating frequency sine function
4


{[2kc,(2k+1)c] (t) sin(24t) + [(2k+1)c,(2k+2)c] (t) sin(48t)},

k=1

with c = 0.2, over the interval [0, 2) using 16,384 sample points (Figure 12a). For
convenience, the function is once again discontinuous; but there also exist continuous
functions with sharp transitions that would pass through these same sampled data. To
obtain a rel. error of 102 , the Coif18 wavelet requires only 301 coefficients (Figure 12b), while the Fourier approximation requires 2,674 (Figure 13b). The errors
pointed out in the Fourier case (Figure 13b) are not seen in the wavelet reconstruction (Figure 12b). Using the same number of terms (310) for the Fourier approximation as for the wavelet series produces a much poorer reconstruction: Its rel. error is

Efficient Signal Transmission

413

Figure 12. a. Original signal. b. Coif18 wavelet series with 310 terms.

Figure 13. a. Fourier approximation with 301 terms for signal in Figure 12a. b. Fourier approximation
with 2,674 terms. The arrows indicate some prominent defects in the Fourier approximation.

414

The UMAP Journal

24.4 (2003)

8.5102 , which is 8.5 times as large as the rel. error for the 301-term Coif18 wavelet
series. Considering that the total number of signal values is 16,384, the 301 nonzero
coefficients for the Coif18 reconstruction represents a significant compression of the
signal.
The two compression examples in this section illustrate some of the variety of signal data that can be compressed using wavelets. For step-function data, which are
locally constant, the Haar wavelets performed wellbetter than a Fourier approximation, and also better than Coif18 wavelets (compare Figures 10b and 10d). This is due
to the step nature of the Haar system. For more smoothly varying data, as in samples
from speech and music, the Coif18 wavelets give excellent compression, clearly superior to Fourier approximations. This was illustrated with a discretely sampled signal
with sharp transitions.
We have seen how wavelets can often allow us to compress our data quite well.
Let us carefully reexamine the properties of wavelets that allowed us to do this. By the
definition of a wavelet, the basis functions m,n (t) = 2m/2 (2m t n) can zoom in
on particular areas of a signal. As m increases, the supports of these basis functions
decrease in size, and varying n allows us to move along to particular areas to examine

the signal. This allows the wavelet coefficients nm = f (t)m,n (t) dt to pick
up information about local behaviors of the function f (t). The vanishing moment
conditions help us to compress the data from smoothly varying signals because many
of the wavelet coefficients nm are either zero or very small. For different types of
signals, different wavelet systems can be employed. In particular, for step functions,
the Haar wavelet system performs very well. While for smoother functions, a smoother
wavelet system like Coif18 works much better. The Coif18 system combines a large
number of vanishing moments for both wavelets and scaling functions (hence good
compression) with a relatively small support (thus allowing for rapid computation).
The Coif18 system is just one of many wavelet systems that can be employed for
compression. Some compression schemes make use of multiple systems, choosing
the best system for a given signal (and transmitting a small amount of overhead to
the receiver to identify which system was used). Besides wavelet series, localized
Fourier seriesFourier expansions restricted to smooth time-windowed portions of
the signalhave been found to be especially effective for compressing audio signals
[Malvar 1992]. More details on data compression can be found in Mallat [1998],
Vetterli and Kovacevic [1995], and Walker [1999].

Conclusion
We have seen how a PCM signal can carry the bits of a digital signal by carrying
the 1s of a binary code with a different frequency than the 0s. By using more frequencies, this PCM signal can carry large numbers of digital signals simultaneously. PCM
signals are also highly resistant to corruption by additive noise.
We next examined three ways to represent a function with such digital codes. The
codes came from the coefficients for a discrete Fourier transform, a discrete Haar
wavelet series, and a discrete Coif18 wavelet series. We saw how wavelets allowed

Efficient Signal Transmission

415

us to compress signals better so that we needed less data to represent a function.


Several of the references would be appropriate reading for undergraduate students.
Students may find Benson [1993] interesting if they wish to read how several radio
signals can be transmitted simultaneously. Walker [1988] and Kelly [2000] give the
reader more of the theory and applications of Fourier series. Finally, the reader may
wish to learn more about wavelets and their uses by consulting Walker [2000; 2002],
Frazier [1999], Walker [1999], and Walker [1997].

References
Boylstad, R.L. 1994. Introductory Circuit Analysis. 9th ed. Upper Saddle River, NJ:
Prentice-Hall,
Benson, C. 1993. How to tune a radio. In Applications of Calculus, vol. 3: Resources
for Calculus, edited by Philip D. Straffin, 126134. Washington, DC: Mathematical Association of America.
Burrus, C.S., R.H. Gopinath, and H. Guo. 1998. Introduction to Wavelets and Wavelet
Transforms, A Primer. Englewood Cliffs, NJ: Prentice-Hall.
Catala, P. (1997). Pulse code modulation. Chap. 23 of The Communications Handbook. Boca Raton, FL: CRC Press.
Daubechies. I. 1988. Orthonormal bases of compactly supported wavelets. Communications on Pure and Applied Mathematics 51: 909996.
. 1992. Ten Lectures on Wavelets. Philadelphia, PA: SIAM.
Floyd, T.L. 2001. Electronics Fundamentals. 5th ed. Upper Saddle River, NJ: PrenticeHall.
Frazier, M. 1999. An Introduction to Wavelets through Linear Algebra. New York:
Springer-Verlag.
Higgins, J.R. 1985. Five short stories about the cardinal series. Bulletin of the American Mathematical Society, New Series 12: 4589.
Kelly, S. 2000. Using the Shannon sampling theorem to design compact discs. The
UMAP Journal 21(2): 157166.
Mallat, S. 1988. Multiresolution approximations and wavelet orthonormal bases of
L2 (R). Transactions of the American Mathematical Society 315(1): 6987.
. 1998. A Wavelet Tour of Signal Processing. New York: Academic Press.
Malvar, H.S. 1992. Signal Processing with Lapped Transforms. Norwood, MA: Artech
House.
Resnikoff, H.L., and R.O. Wells. 1998. Wavelet Analysis. The Scalable Structure of
Information. New York: Springer-Verlag.

416

The UMAP Journal

24.4 (2003)

Vetterli, M., and J. Kovacevic. 1995. Wavelets and Subband Coding. Englewood
Cliffs, NJ: Prentice-Hall.
Walker, J. 1988. Fourier Analysis. New York: Oxford University Press.
. 1996. Fast Fourier Transforms. Boca Raton, FL: CRC Press.
. 1997. Fourier analysis and wavelet analysis. Notices of the American
Mathematical Society, 44(6): 658670.
. 1999. A Primer on Wavelets and their Scientific Applications. Boca Raton,
FL: CRC Press.
. 2000, 2002. FAWAV: A Fourier/Wavelet Analyzer. Freeware for Windows.
http://www.uwec.edu/walkerjs/ .

About the Authors


Susan Kelly received her Ph.D. in mathematics
in 1992 from Washington University in St. Louis,
with a thesis in wavelet theory under Richard Rochberg
and Mitchell Taibleson. She is a professor at the
University of WisconsinLa Crosse. Her hobbies
include watercolor and oil painting as well as skiing.
She enjoys spending time with her husband and her
son Kyle.

James S. Walker received his B.S. from the State University of New York at Buffalo (1975), his M.S. from the
University of Illinois at UrbanaChampaign (1977), and his
D.A. from the University of IllinoisChicago (1982). He is
a professor of mathematics at the University of Wisconsin
Eau Claire. He has published papers on Fourier analysis,
complex variables, and wavelet theory. He is the author of
three books on Fourier analysis, FFTs, and wavelet analysis.

UMAP Module 792


Modules in
Undergraduate
Mathematics
and Its
Applications

The Spread of
Forest Fires
Emily E. Puckette

Published in
cooperation with
The Society for
Industrial and
Applied Mathematics,
The Mathematical
Association of America,
The National Council
of Teachers of
Mathematics,
The American
Mathematical
Association of
Two-Year Colleges,
The Institute for
Operations Research and
the Management
Sciences, and
The American
Statistical Association.

Application Field: Biology, Forestry


COMAP, Inc., Suite 210, 57 Bedford Street, Lexington, MA 02420 (781) 8627878

418

The UMAP Journal

24.4 (2003)

INTERMODULAR DESCRIPTION SHEET:

UMAP Unit 792

TITLE:

The Spread of Forest Fires

AUTHOR:

Emily E. Puckette
Department of Mathematics and Computer Science
Sewanee: The University of the South
735 University Ave.
Sewanee, TN 37383-1000
eep@sewanee.edu

MATHEMATICAL FIELD:

Probability, Calculus

APPLICATION FIELD:

Biology, Forestry

TARGET AUDIENCE:

Students with a background in innite series, e.g., from


second-semester calculus.

ABSTRACT:

We create a simple discrete probabilistic model for


spread of a forest re. We examine the conditions for
which the re will either die out or spread indenitely,
identifying and bounding a critical value for the probability of transmission of the re to an immediately adjacent location.

PREREQUISITES:

Summation of geometric series, elementary probability, intuitive limits; for the Appendix, derivatives and
limits of rational functions.

The UMAP Journal 24 (4) (2003) 417434.

c
Copyright
2003 by COMAP, Inc. All rights reserved.
Permission to make digital or hard copies of part or all of this work for personal or classroom use
is granted without fee provided that copies are not made or distributed for prot or commercial
advantage and that copies bear this notice. Abstracting with credit is permitted, but copyrights
for components of this work owned by others than COMAP must be honored. To copy otherwise,
to republish, to post on servers, or to redistribute to lists requires prior permission from COMAP.

COMAP, Inc., Suite 210, 57 Bedford Street, Lexington, MA 02420


(800) 77-COMAP = (800) 772-6627, or (781) 862-7878; http://www.comap.com

The Spread of Forest Fires

419

The Spread of Forest Fires


Emily E. Puckette
Dept. of Mathematics and Computer Science
Sewanee: The University of the South
735 University Ave.
Sewanee, TN 37383-1000
eep@sewanee.edu

Table of Contents
1. INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2. FOREST FIRE MODEL . . . . . . . . . . . . . . . . . . . . . . . . . .

3. SPREAD OF THE FOREST FIRE . . . . . . . . . . . . . . . . . . . . . .

4. LOWER BOUND FOR pc . . . . . . . . . . . . . . . . . . . . . . . . .

5. UPPER BOUND FOR pc . . . . . . . . . . . . . . . . . . . . . . . . . .

6. CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

7. AREAS FOR FURTHER EXPLORATION . . . . . . . . . . . . . . . . . .

8. SOLUTIONS TO SELECTED EXERCISES . . . . . . . . . . . . . . . . . . 11


9. APPENDIX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
ABOUT THE AUTHOR . . . . . . . . . . . . . . . . . . . . . . . . . . 13

420

The UMAP Journal

24.4 (2003)

MODULES AND MONOGRAPHS IN UNDERGRADUATE


MATHEMATICS AND ITS APPLICATIONS (UMAP) PROJECT
The goal of UMAP is to develop, through a community of users and developers, a system of instructional modules in undergraduate mathematics and
its applications, to be used to supplement existing courses and from which
complete courses may eventually be built.
The Project was guided by a National Advisory Board of mathematicians,
scientists, and educators. UMAP was funded by a grant from the National
Science Foundation and now is supported by the Consortium for Mathematics and Its Applications (COMAP), Inc., a nonprot corporation engaged in
research and development in mathematics education.

Paul J. Campbell
Solomon Garfunkel

Editor
Executive Director, COMAP

The Spread of Forest Fires

1.

421

Introduction

What does research in mathematics look like?


What do mathematicians do, exactly?
Is there any thing new going on in mathematics?
Such questions are exciting, because students asking them are curious about
mathematics, and frustrating, because the students often do not have enough
background in mathematics to tackle a research problem. The challenge to the
instructor is to respond with an easily understood research question whose
solution is within the grasp of a freshman or sophomore. An added bonus
would be to convey the creative mix of ideas that is often crucial in a solution.
The project that follows is one possible response. It begins with a very
approachable model for the spread of a forest re. The question of interest is
whether the re will die out on its own, and this question leads to a hunt for a
threshold value for the probability of further transmission of the re. With use
of the formula for the sum of a geometric series plus a minimal amount of probability, a partial answer is attainable: some bounds on the critical probability
of spread of the re.
The main ideas in this Module have been used
in senior comprehensive projects,
as a basis for talks given to undergraduates with a second-semester calculus
background, and
as an independent study for a motivated sophomore.
As organized here, the Module is written as an independent project for a student
or a group of students. The time frame for the project is 24 weeks.

2.

Forest Fire Model

The model is constructed on the Z2 lattice, where each lattice point X =


(x, y), x, y Z, is the site of a tree. The tree at the origin is struck by lightning
and set ablaze. From any tree, the re can spread in one step to any one of its
four nearest neighbors, i.e., to any of the trees at (1, 0), (0, 1), (1, 0) or (0, 1),
independent of one another, each with transmission probability p, 0 p 1.
We keep track of the spread of the re using a discrete time clock. Thus,
at time n = 0, the tree at the origin is on re and all other trees in the forest
are unburnt. At time n = 1, the re may have spread to one or more nearest
neighbors of the tree at the origin. Also at time n = 1, the tree at the origin
is completely burnt and therefore cannot be set on re again nor transmit re
further.

422

UMAP/ILAP Modules 200304

We formalize the re spread by assigning to each lattice site X a state at time


n, n = 0, 1, 2, . . . The three possible states for a tree are U (unburnt), F (on re),
and B (burnt). The function Sn (X) gives the state of the tree at site X at time n.
Exercise
1. The spread of the re is determined by the following rules. Translate them
into words; for example, Sn (X) = B translates to the tree at site X at time
n is burnt.
a) If Sn (X) = B, then Sn+1 (X) = B.
b) If Sn (X) = F , then Sn+1 (X) = B.
c) If Sn (X) = U , then Sn+1 (X) can be either U or F , depending on the
state of the neighboring trees and the transmission of the re.
Case c) requires some exploration. A burning tree spreads the re to neighboring trees independently of other burning trees. If events A and B are independent, then the probability that both events A and B occur, P (AB), is equal
to P (A) P (B), the product of the probabilities of each separate event. In other
words, the proportion of the time that event A occurs is the same when event
B occurs as when B does not occur.
With this precise denition of independence, consider the unburnt tree at
site X with neighboring trees at X1 , X2 , X3 , and X4 , as in Figure 1.

X2
X3

X1

X4
Figure 1. An unburnt tree at X with neighboring trees.

Exercises
2. If Sn (X1 ) = F , Sn (X2 ) = F , Sn (X3 ) = B, and Sn (X4 ) = U , the tree at X
will remain unburnt at time n + 1 only if the re at sites X1 and X2 do not
spread to X. The probability of not spreading the re from a burning tree
to a neighbor is 1 p.
Let A be the event that site X1 s re does not spread to site X, and
let B be the event that site X2 s re does not spread to site X. Using the

The Spread of Forest Fires

423

independence of the re spread from site to site, the probability that site X
remains in state U at time n + 1 is
P (A B) = P (A) P (B) = (1 p)2 .
What is the probability that the site X switches to state F at time n + 1?
That is, what is the probability that the tree at X is on re at time n + 1?
3. Consider this assignment of states to sites: Sn (X1 ) = F , Sn (X2 ) = B,
Sn (X3 ) = F , and Sn (X4 ) = F .
a) If Sn (X) = U , what is the probability that Sn+1 (X) = U ?
b) Again, if Sn (X) = U , what is the probability that Sn+1 (X) = F ?
4. Consider the case where at time n exactly i neighboring trees are on re,
i = 0, 1, . . . , 4.
a) If Sn (X) = U , what is the probability that Sn+1 (X) = U ?
b) Again, if Sn (X) = U , what is the probability that Sn+1 (X) = F ?

3.

Spread of the Forest Fire

The spread of the re depends on the value of the transmission probability


p. If p is small, the re will die out eventually. If p is large, then there is some
chance that the re will spread innitely, that is, arbitrarily far from the initial
lightning strike.
The Big Question: How large does p have to be for there to be a chance for
indenite spread of the re?
Essentially, we are hunting for a threshold value or critical value pc . For p
[0, pc ), the forest re will die out. For p (pc , 1], the re has a chance of innite
spread. However, unless p = 1, it is not certain that there will be innite spread
of the re.
The model plays out differently depending on the value of p; the value
pc marks a phase transition in the model. Phase transitions abound in natural
phenomena, too, as in the change of water from a liquid state to a solid state,
which (for pure water at sea level) occurs at a critical temperature tc = 0 C.
Exercise
5. Computer simulation can provide some insight into the critical value pc .
Write a computer program that starts with one tree burning at the center of
a 100 100 lattice and shows the states of the trees at time n using the rules
for spread established in the previous section. Explore the re spread with
different values of the transmission probability p; to begin, test p = 0.2,
p = 0.4, p = 0.6, and p = 0.8. Look at the states of the entire lattice at
3

424

UMAP/ILAP Modules 200304

various times, for example, at n = 20, n = 40, and n = 60, to see how the
re is spreading over time. For each value of p, run the program several
times to observe any trends in how the re spreads. You may also want to
increase the size of the lattice to explore further. What do you notice about
the spread of the re? For which values of p does the re appear to die out?
For which values does the re seem to continue spreading?
For convenience, let E be the event that the re spreads innitely. Hence
for p < pc , the probability of E is 0, while if p > pc , the probability of E is
greater than 0. The critical probability is therefore dened to be
pc = inf{p | P (E ) > 0}
where inf denotes the greatest lower bound of the set.

4.

Lower Bound for pc

To study innite spread of the re, we start by considering how the re


spreads from the origin to another site at a nite distance away. As the re
spreads across the lattice, it creates burnt paths through the forest. We consider
one such path.
Exercises
6. a) Plot all of the Z2 lattice points (x, y) where 4 x, y 4. At time n = 0,
the tree at the origin is on re. Suppose that the re spreads to one of the
neighboring trees. Pick one neighboring tree, at site X, say. How many
options are there for choice of X? On your plot, draw the rst step of
the res path from the origin to X.
b) For the next step in the spread of the re, pick a tree neighboring X; call
the site Y . How many options are there for choice of Y ? On your plot,
draw the second step of the res spread.
c) For the third step, pick a tree neighboring Y . How many options are
there? On your plot, draw the next step of the res spread.
Because the re cannot revisit a burnt tree, it maps out a self-avoiding path,
meaning that the path cannot return to a site it has already visited.
Exercise
7. a) How many self-avoiding paths with a length of 2 steps are there on this
lattice? How many with three steps?
b) Are there 4 3 3 3 self-avoiding paths with a length of 4 steps? Sketch
several paths and think about the question.

The Spread of Forest Fires

425

Depending on where the re is at time n = 3, it might not have three options


of sites to move to at time n = 4. For example, suppose that the rst step from
the origin is east, the second step is north, and the third step is west. Then the
fourth step must be either north or west.
Counting the number of self-avoiding paths with n steps is a difcult problem, but we can at least establish an upper bound on the number of such paths:
Number of self-avoiding paths of n steps 4 3n1 .
Then the probability that the re spreads along a self-avoiding path of n steps
is
(Number of n-step self-avoiding paths)(Probability of re spread n times)
= (Number of self-avoiding paths of n steps) pn (4 3n1 )pn .
Now consider innite spread of the re; re must spread along a selfavoiding path of innite length. So let us consider re spreading along a
self-avoiding path with a large length.
Let Q be the probability of re spreading along a self-avoiding path with N
or more steps, where N is large. Then
Q = P (re spreading along a self-avoiding path of length N )


P (re spreading along a self-avoiding path of length n)

n=N

(4 3n1 )pn

n=N

= 4p

(3p)n1 .

n=N

Exercise
8. If p < 13 , then 0 3p < 1, and the sum in Q is the tail of a convergent
geometric series. Use the closed form ofthe series to simplify the upper

bound of Q. Hint: The geometric series n=0 axn converges to a/(1 x)


for 1 < x < 1 and a R. For more assistance, see the Appendix.
For innite spread of the forest re, we are interested in the limiting probability of re spreading along a self-avoiding path of length N , that is, the limit
of Q as N .
Exercises
9. Assume that p < 13 . Use the simplied upper bound of Q to nd an upper
bound for limN Q.

426

UMAP/ILAP Modules 200304

10. Note that


P (E ) = lim Q,
N

so we can conclude that if p < 13 , then the probability of innite re spread


is 0. What then can be said about the value of the critical probability pc ?

5.

Upper Bound for pc

To obtain an upper bound for pc , we consider the spread of a re at time n = 0


from a set I of M initial adjacent sites in a row: (0, 0), (1, 0), (2, 0), . . . , (M, 0).
The re spreads along self-avoiding paths; so let J be the set of sites X such
that there is a self-avoiding path from a site in I to X. That is, J consists of all
sites that are burned by the res spread; note that I J. See Figure 2 for an
illustration.


0


M


 - site in J
Figure 2. Illustration of a re starting from a row of trees from 0 to M .

For each site Y in J, construct a closed unit square centered at Y . So if


Y = (x, y), then the unit square has vertices at (x 12 , y 12 ). Let A be the
region of the plane covered by these squares, and let A have boundary B. See
Figure 3 for an illustration.
If there is indeed a boundary around all the sites that are burnt, then the
re did not spread innitely from the initial sites. So our attention turns to the
probability of the existence of such a boundary B.
In constructing the unit squares around sites in J, we have effectively created
a dual lattice W2 , where


W2 = (u, v) : u = x + 12 , v = y + 12 for all (x, y) Z2 .
6

The Spread of Forest Fires


0


M


427

boundary B
Figure 3. The boundary B of the region A corresponding to the re in Figure 2.

The boundary B connects lattice points in this dual lattice.


The spread of the re in the Z2 lattice induces a conguration in the W2
lattice in the following way. In the Z2 lattice, let X and Y be neighboring sites.
We call the segment of the lattice between the sites an edge. If the re has spread
from one site to a neighboring site, we say the edge is open; otherwise, the edge
is closed. Thus, an edge in the Z2 lattice is open with probability p, and as the
model runs its course, each edge in Z2 is classied as either open or closed.
Every edge in the Z2 lattice crosses an edge in the W2 lattice. If the edge
is open in Z2 , the corresponding edge is declared to be open in the W2 lattice.
In our example (Figure 3), the edge in Z2 from the origin to the point (0, 1) is
open because the re has spread from the origin to this site. Therefore, the edge
in W2 from ( 12 , 12 ) to ( 12 , 12 ) is also open. Thus a conguration in W2 of open
and closed edges is determined by the conguration of open and closed edges
in Z2 .
The boundary B consists of closed edges in W2 . Furthermore, B is a selfavoiding path until it closes the loop around the burnt area in its last step.
Exercises
11. What is the smallest possible number N of edges in B? Consider the minimum size of the set J, and determine the minimum number of edges in W2
that would be required to bound J.
12. Consider a boundary B of some length k, where k N ; it must be a path
formed of k closed edges. What is the probability that all k edges are closed?

428

UMAP/ILAP Modules 200304

How many boundaries of length k exist? Since a boundary is a self-avoiding


path, the question is difcult to answer precisely. However, we can establish
an upper bound for the number of boundaries of length k. Since the initial set
I is on the horizontal axis of Z2 , a boundary of length k must contain a vertical
edge between W2 lattice points of the form (j + 12 , 12 ) and (j + 12 , 12 ) for some
integer j where 0 j < k. Thus, there are at most k options for an edge of this
form; and moving in a self-avoiding fashion from that edge onward, there are
at most 3k1 options for the following k 1 steps. Therefore, the number of
boundaries of length k is less than or equal to k 3k1 .
Let R be the probability of the existence of a closed boundary B around
set J. Then we have
R = P (closed boundary B exists)


P (closed boundary B of length k exists)
=

k=N

(k 3k1 )(1 p)k

k=N

= (1 p)

k[3(1 p)]k1 ,

k=N

where N = 2M + 4 is the minimum number of edges in the boundary.


Exercise
sum is the tail of a convergent series.
13. If p > 23 , then 0 3(1 p) < 1, and the

(The ratio test shows that the series k=1 k[3(1 p)]k1 is convergent, and
Exercise A2 calculates what it converges to; but here we are concerned just
with the tail of the series, the sum from the N th term on.) What happens to
R if N is increased? Hint: See Exercise A2 in the Appendix.
Assume that p > 23 . If N = 2M + 4 is large enough, then R < 1. But if the
probability of the existence of a closed boundary B around set J is less than 1,
then the probability of no closed boundary around set J is greater than 0. That
implies that at least one of the trees in the initial set I was the starting point
of an innite open self-avoiding path. Since it is equally likely for any one of
these trees to be the starting point of such a path, we have

The Spread of Forest Fires

429

0 < P (closed boundary B does not exist)


= P (set I generates innite re spread)


P innite open self-avoiding path starting at (0,0) exists


+ P innite open self-avoiding path starting at (1,0) exists
+ ...


+ P innite open self-avoiding path starting at (M,0) exists


= (M + 1) P innite open self-avoiding path starting at (0,0) exists
= (M + 1) P (E ).
Exercise
14. So we can conclude that if p > 23 , then the probability of innite re spread
P (E ) is greater than 0. What then can be said about the value of the critical
probability pc ?

6.

Conclusions
We now have bounds on the critical probability pc :
1
2
pc .
3
3

You should review your computer simulation results to see how those results
agree. The next mathematical step would be to improve the techniques so as
to narrow the bounds of pc , and ultimately nd the exact value of the critical
probability. The proof that
pc actually equals

1
2

was rst published by Kesten [1980], and other proofs have subsequently appeared (e.g., Grimmett [1989]).

7.

Areas for Further Exploration

There are other interesting questions besides innite spread. For example,
for transmission probabilities below the critical probability, we know that the
re will die out.
How far will it spread on average before it dies out?
What can be determined about the shape of the burnt region? Is it circular,
or does it exhibit some other pattern?
9

430

UMAP/ILAP Modules 200304

The answers depend on the size of the transmission probability p.


We could explore variations on the model. The lattice for the model is
square, with each tree having four nearest neighbors. However, we could also
model using a lattice that is triangular or hexagonal, with each tree having three
or six nearest neighbors. How can the methods for nding upper and lower
bounds of pc be adjusted for a different lattice? Do the bounds change?
In the model, we held the transmission probability p xed. However, it is
reasonable to consider that the probability of spread may vary. In particular,
a constant wind would make it more likely for a re to spread in one direction. Consider a constant wind from the south that makes the probability of
re spreading to the north more likely than toward any other direction. Use
computer simulations to explore this with the probability of northward spread
above the threshold value pc = 12 and the probability of eastern, western and
southern spread below the threshold value. What do you predict will happen?
Another adjustment is to consider variable winds. A wind coming from
the south with variable speed can be worked into the computer model as a
random choice from several transmission probability values; the stronger the
wind, the higher the probability of spread to the north. One could make it
more complicated by having random winds from any direction at any time, or
stronger winds as the re grows.
There are many other models with similar structures and questions. For
example, the model for the spread of forest re can also be used to study the
spread through a population of an immunity-granting illness such as measles
or chicken pox (e.g., Durrett [1988a]). A closely related model is the percolation
model. The model is set on the Z2 lattice with edges between sites open with
probability p, and a central question is whether the origin is in an innite open
cluster. That is, is there is an open path for uid (such as water or oil) to move
from the origin to innity?
This project is an entry point into the study of interacting particle systems
and cellular automata. Those models all follow simple rules to determine the
states of sites, and yet they produce complex behaviors for study. Consider the
most famous cellular automaton of all, John Conways Game of Life. In this
model, sites have only two states, alive and dead, and at each time step, sites
change states depending on the states of their nearest neighbors. Even with
these simplest of transition rules, interesting evolutions occur (see Gardner
[1970; 1983] and computer implementations such as Schaffhauser [1996]).

10

The Spread of Forest Fires

8.

431

Solutions to Selected Exercises

1. a) If the tree at site X at time n is burnt, then the tree at that site is burnt at
the following time n + 1.
b) If the tree at site X at time n is on re, then the tree at that site is burnt
at the following time n + 1.
c) If the tree at site X at time n is unburnt, then the tree at that site is either
unburnt or on re at the following time n + 1.
2. The probability that the tree at site X is on re at time n + 1 is 1 (1 p)2 .
3. a) If Sn (X) = U , the probability that Sn+1 (X) = U is (1 p)3 .
b) If Sn (X) = U , the probability that Sn+1 (X) = U is 1 (1 p)3 .
4. a) If Sn (X) = U , the probability that Sn+1 (X) = U is (1 p)i .
b) If Sn (X) = U , the probability that Sn+1 (X) = U is 1 (1 p)i .
6. a) There are 4 options for the choice of X.
b) There are 3 options for the choice of Y .
c) There are 3 options for the third step.
7. There are 4 3 = 12 self-avoiding paths of length 2 and 4 3 3 = 36 selfavoiding paths of length 3.
a) There are fewer than 4 3 3 3 self-avoiding paths of length 4.
8. Let x = 3p. Then Q 4p

(3p)n1 = 4p

n=N

(3p)N 1
.
1 3p

(3p)N 1
= 0.
N 1 3p

9. lim Q 4p lim
N

10. Since pc = inf{p : P (E ) > 0}, pc cannot be less than 13 . Thus pc 13 .


11. The minimum set is I itself, and for this set, N = 2(M + 1) + 2 = 2M + 4.
12. The probability of k edges being closed is (1 p)k by independence.
13. If N is increased, the probability R decreases.
14. Since pc = inf{p : P (E ) > 0}, pc cannot be greater than 23 . Thus pc 23 .

11

432

UMAP/ILAP Modules 200304

9.

Appendix

We give more details on the geometric series used in several exercises. Let
x be a real number. For some integer M , dene the sum SM to be
2

SM = 1 + x + x + x + + x

M


xk .

k=0

Exercises
A1. Consider the derivative of SM :
M

dSM
= 0 + 1 + 2x + 3x2 + + M xM 1 =
n xn1 .
dx
n=1

Take the derivative of the closed form of SM to nd another expression of


the derivative:

d
1 xM +1
dSM
=
= .
dx
dx
1x
dSM
. Does this limit exist for 1 < x < 1?
M dx

Let S  = lim

A2. For 1 < x < 1, let N = M + 1 and dene the tail of the series
TN


d SM
=
=S
n xn1 .
dx


n=N

What is the limit of TN as N ?


Solutions
A1.

A2.

dSM
(M + 1)xM (1 x) + (1 xM +1 )
SM
(M + 1)xM
=
+
.
=
dx
(1 x)2
1x
1x
For 1 < x < 1, the limit does exist, and S  = 1/(1 x)2 .
lim TN = 0.

12

The Spread of Forest Fires

433

References
Durrett, R. 1988a. Crabgrass, measles, and gypsy moths: An introduction to
interacting particle systems. Mathematical Intelligencer 10: 3747.
. 1988b. Lecture Notes on Particle Systems and Percolation. CA: Wadsworth
and Brooks/Cole.
Gardner, Martin. 1970. The fantastic combinations of John Conways new solitaire game Life. Scientic American 223 (4): 120ff. 1983. Reprinted with
substantial additional commentary in Wheels, Life and Other Mathematical
Amusements, 214257. NY: W.H. Freeman.
Grimmett, G. 1989. Percolation. NY: Springer-Verlag.
Kesten, H. 1980. The critical probability of bond percolation on the square
lattice equals 12 . Communications in Mathematical Physics 74: 4159.
Liggett, T.M. 1985. Interacting Particle Systems. NY: Springer-Verlag.
Schaffhauser, Christoph. 1996. Game of Life. Version 1.7.0. Computer program
for PowerPC and Motorola 680x0 Macintosh computers. Available at http:
//www.datacomm.ch/darkeagl/ .

About the Author


Emily Puckette is an associate professor of
mathematics at the University of the South in Sewanee, TN. She earned her B.A. in mathematics
at Smith College and her Ph.D. at Duke University, focusing on critical probabilities and random walks. One of her favorite extracurricular
activities is to go walking in the woods, pondering probabilities.

13

434

UMAP/ILAP Modules 200304

14

Getting the Salt Out

N T E R D I S C I P L I N A R Y

I V E L Y

P P L I C A T I O N S

435

R O J E C T

AUTHORS:
Ethan Berkove
(Mathematics)
berkovee@lafayette.edu

Getting the Salt Out

Thomas Hill
(Mathematics)

MATHEMATICS CLASSIFICATIONS:
Calculus, Differential Equations

Scott Moor
(Chemical Engineering)

DISCIPLINARY CLASSIFICATIONS:
Chemistry, Engineering

Lafayette College
Easton, PA 18042

PREREQUISITE SKILLS:
1. Modeling with an ordinary differential equation with initial
condition.
2. Sketching a solution to a differential equation on its direction
field.
3. Solving an ordinary differential equation by separation of variables.

CONTENTS
1. Setting the Scene
2. Your Job
3. Part 1: Numerical and
Graphical Analyses
4. Part 2: Let's Get Explicit
5. Part 3: What Do You
Recommend?
References
Sample Solution
Notes for the Instructor
About the Authors

PHYSICAL CONCEPTS EXAMINED:


Osmotic pressure, permeability, desalination.
COMPUTING REQUIREMENT:
A program to plot direction fields. (A Mathematica program is
available from the authors.)

The UMAP Journal 24 (4) 435-449. Copyright 2003 by COMAP, Inc. All rights
reserved. Permission to make digital or hard copies of part or all of this work for personal or classroom use is granted without fee provided that copies are not made or
distributed for profit or commercial advantage and that copies bear this notice.
Abstracting with credit is permitted, but copyrights for components of this work
owned by others than COMAP must be honored. To copy otherwise, to republish, to
post on servers, or to redistribute to lists requires prior permission from COMAP.

436

The UMAP Journal

24.4 (2003)

Contents
1. Setting the Scene
2. Your Job
3. Part 1: Numerical and Graphical Analyses
4. Part 2: Lets Get Explicit
5. Part 3: What Do You Recommend?
References
Sample Solution
Notes for the Instructor
About the Authors

1.

Setting the Scene

When two water (or other solvent) volumes are separated by a semi-permeable
membrane, water will ow from the side of low solute concentration to the side
of high solute concentration. This is known as osmosis (see Figure 1). The ow
of solvent across the membrane may be stopped, or even reversed, by applying
external pressure on the side of higher solute concentration. This process is
called reverse osmosis (see Figure 2).

Figure 1. Osmosis.

Figure 2. Reverse osmosis.

We will use vant Hoffs equation to model osmosis. Jacobus Henricus


vant Hoff (18521911) determined that the osmotic pressure is given by the
equation
= cRT,
where
c is the molar solute concentration,
R is the universal gas constant, and
T

is the absolute temperature.

Notice that c = n/V , where n is the number of moles of solute and V is the
volume of solution. Although vant Hoffs equation looks like a restatement
of the ideal gas law, P V = nRT , it is special because it is being applied to a

Getting the Salt Out

437

liquid rather than to a gas. This equation was a signicant development; in


1901 vant Hoff received the Nobel Prize in Chemistry in recognition of the
extraordinary services he has rendered by the discovery of the laws of chemical
dynamics and osmotic pressure in solutions [Nobel Foundation 2000].
Consider the case in Figure 2, where an external pressure, P , is applied to
the side of the membrane with higher solute concentration. The resulting ux
through the membrane is proportional to the difference between the osmotic
pressure and the applied pressure. That is, if x represents the volume of water
that has been extracted from the solution, then
dx
= A(P ),
dt

(1)

where is the water permeability constant andA is the membrane area perpendicular to the ow.
It is important to note that is not a constant here. If the initial volume of
solution is V , and if x units of water have been extracted from the solution at
time t, then
n
RT.
(t) =
V x
Substituting this expression into (1), we get the desalination differential equation,

n
cV RT
dx
= A P
RT = A P
.
(2)
dt
V x
V x

2.

Your Job

Clearwater, Inc., is planning to build a new, portable water purier to remove salt from seawater. They have designed a machine, shown below in
Figure 3; but before they go to the expense of building a prototype, they want
you to perform a theoretical analysis of their design.
Your task is to analyze how the values of the design parameters for the
desalination machine will affect the performance of the machine. Keeping in
mind that the machine must be portable:
How should the membrane coefcient , the membrane size A, the volume V of
the desalination chamber, and the applied pressure P be chosen to produce a
machine that is both efcient and economical?
Clearwater is counting on you to design a good product, so dont let them
down!

438

The UMAP Journal

24.4 (2003)

Salt Water
Applied Load

Revese
Osmosis Membrane
Reverse
Osmosis

Desalinated Water

Figure 3. Clearwaters desalination machine.

3.

Part 1: Numerical and Graphical


Analyses

Requirement 1
Use direction elds and Eulers method (of following along the directions
in a direction eld) to sketch solutions of (2). Of course, to do that you will
need to assign values to the various constants and parameters in the equation.
For the physical constants in the problem, use the values
moles
,
L
Lbar
,
R = 0.082
mole K
T = 293 K.
c = 0.103

For the design parameters, begin with


= 0.1

m3
,
m2 daybar

A = 1.5 m2 ,
P = 15 bars,
V = 4 L.
Plot a direction eld for the desalination differential equation where t (measured in days) goes from 0 to 5 and x (measured in L) goes from 0 to 4. Also,

Getting the Salt Out

439

plot the solution curve for x that passes through the point (0, 0).
According to your direction eld and solution, how much fresh water can
the machine extract from 4 L of sea water?
Approximately how long does it take to extract 2 L of fresh water?
As time goes by, what happens to the rate at which the machine produces
fresh water? Does this make sense to you? Can you think of a physical reason
why this would happen?

Requirement 2
Suppose that we change the capacity of the machine. For example, suppose
that the initial volume of brine is 6 L instead of 4. How much fresh water
can we extract? How long does it take to extract 2 L of fresh water? What
happens if we begin with 8 L of brine? How sensitive is the time required to
extract 2 L of fresh water to the initial volume of sea water in the machine?
What relationship seems to exist between the initial volume of sea water and
the total amount of fresh water that can be extracted from it? The graphs in
Figure 4 show examples of proportionality relationships that you might look
for in your investigations.

Figure 4. Three proportionality relationships: v u, v u2 , and v 1/u.

Requirement 3
Now x the value of V at 6 L and investigate how the performance of the
machine depends on the other design parameters. What effect does increasing
or decreasing A have on the amount of fresh water that we can extract from
a given volume of brine? On the time required to extract 2 L of fresh water?
Answer the same questions for and P . Of the three variables, which one,
when varied, causes the greatest change in the amount of fresh water extracted?
On the time required to extract 2 L of fresh water?

Requirement 4
What is the equilibrium solution of (2)that is, the solution when dx/dt =
0? Discuss how that solution depends on the design parameters and compare
your conclusions to the observations you made in the earlier parts above.

440

4.

The UMAP Journal

24.4 (2003)

Part 2: Lets Get Explicit

Requirement 5
Use separation of variables to obtain the solution of (2) that satises the
initial condition that x = 0 when t = 0. Hint: Solve for t as a function of x.

Requirement 6
Use the solution of the differential equation to determine how the performance of the Clearwater desalination machine depends on the values of the
parameters , A, and P . Do your results agree with the conclusions that you
drew earlier using direction elds?

5.

Part 3: What Do You Recommend?

Requirement 7
Prepare a report for the engineers working on the desalination project at
Clearwater, Inc. An important issue is how quickly the proposed machine will
be able to produce 2 L of fresh water. Your report should include an analysis
of how this time depends on each of the design parameters , A, and P , and
on V .
Clearwaters marketing people have reported that, if their machine is to
be competitive in the marketplace, it must be able to produce at least 2 L of
fresh water per day. If the machine is to be portable, it can handle at most
8 L of seawater at a time. Furthermore, because the walls of the pressure
vessel are quite thin, the upper limit on the applied pressure, P , is 30 bars.
Filters for this machine are extremely expensive; consequently, the area of the
lter is to be no greater than 1.2 m2 . Finally, the best lter available has a
permeability coefcient of = 0.08. Is it theoretically possible to build a reverse
osmosis desalination machine that meets all of these specications? What is
the approximate size of the smallest lter that can be used in this machine to
meet the specications?
To nish your report, give your recommendation for values of design parameters that yield the best machine. A best machine should be built
with some margin for error, so pick your parameter values with some room to
maneuver (is 510% possible?) if you can.
The engineers who will be reading your report are familiar with vant Hoffs
equation. Therefore, you do not need to derive or explain that equation. However, you do need to provide mathematical explanations and justications for
all of the conclusions in your report. You should include all relevant graphs in
an orderly fashion.

Getting the Salt Out

441

Title: Getting the Salt Out

Instructors Solution
Part 1: Numerical and Graphical Analyses
Requirements 1 and 2
With the given values of the physical constants and design parameters, the
direction eld of (2) and the solution through (0, 0) are as shown in Figure S1.

Figure S1. Direction eld for V = 4 L.

From this graph, it is apparent that limt x(t) 3.4 L. Furthermore, 2 L


of fresh water is obtained in approximately 1.1 days. Fresh water is extracted
at a decreasing rate because as fresh water is extracted, the concentration of
salt in the remaining brine solution increases. To maintain a constant rate of
extraction of fresh water, it would be necessary to increase the applied pressure.
With V = 6 L and the other design parameters unchanged, it is possible to
extract a little more than 5 L of fresh water from the solution, and the rst 2 L of
fresh water can be extracted in just over 1 day. This is illustrated in Figure S2.
For V = 8 L, it is possible to extract 6.8 L of fresh water, but the time for 2 L
is almost unchanged from when the volume was 4 L or 6 L. With additional
experimentation, a student can observe that the total fresh water extractable
from a brine solution is proportional to the initial volume of the solution.

Requirement 3
Graphs such as those in Figure S3 suggest that changing the area of the
membrane has no effect on the amount of fresh water extractable from a brine
solution. Furthermore, the time for 2 L of fresh water is inversely proportional
to the area of the membrane. (In Figures S3S5, the values of the parameters
are = 0.1, A = 1.5 m2 , P = 15 bars, and V = 4 L except as indicated.)

442

The UMAP Journal

24.4 (2003)

Figure S2. Direction eld for V = 6 L.

Figure S3. Effect of increasing the membrane size: A = 1.0, 1.5, 2.0 m2 .

The graphs in Figure S4 lead to the same conclusions about the relationship
between the water permeability constant and the efciency of the machine.

Figure S4. Effect of increasing the water permeability constant: = 0.04, 0.06, 0.08 m/daybar.

The graphs in Figure S5 show that the total amount of fresh water that can
be extracted from a given volume of brine is a (slowly) increasing function of
the applied pressure P . The time required to extract 2 L of fresh water from
a given solution seems to decrease a little faster than 1/P .
These analyses indicate that:
We can get more fresh water from the solution only by increasing either the
volume V or the applied pressure P , and that we get much more for our efforts
by increasing V .
The time to extract 2 L of fresh water from a given volume of brine can be
reduced by increasing any of the parameters , A, and P ; changing P has
the greatest effect, but only marginally.

Getting the Salt Out

443

Figure S5. Effect of increasing the applied pressure: P = 10, 15, 20 bars.

Requirement 4
The equilibrium solution of (2) is
cV RT
=V
x=V
P

cRT
1
P

Therefore, as we predicted, the total amount of fresh water that can be extracted
from a given volume of brine is proportional to the initial volume of the solution.
Notice that, as expected, the amount of water that can be extracted does not
depend on either A or . The equilibrium solution also shows that the amount
of fresh water we can extract increases (gradually) with increasing P .

Part 2: Lets Get Explicit


Requirement 5
We could solve

dx
cV RT
= A P
dt
V x

with the initial condition x = 0 when t = 0 via a computer algebra system, but
in fact separation of variables and the integration can be done by hand:
V x
dx = A dt
P (V x) cV RT
x
t
V u
du =
A dt,
0 P (V u) cV RT
0


cRT
P x
x

= At,
ln 1 +
P
(P )2
cV RT V P



cV RT
P x
1
ln 1 +
t(x) =
x
.
AP
P
cV RT V P

444

The UMAP Journal

24.4 (2003)

Requirement 6
The solution of the initial value problem shows that t(x), the time required to
extract a specied volume of fresh water from a brine solution, is inversely proportional to both and A. Furthermore, the solution shows that t(x) decreases
a little faster than 1/P , and that there is only a very weak relationship between t(x) and V . The argument of the logarithm approaches 0 as x approaches
V cV RT /P . This is the equilibrium solution that we found before.
The graph of t(x) with = 0.1, A = 1.5, P = 15, and V = 6 is given in
Figure S6; it is the reection of the solution curve that we saw in Figure S1
about the line x = t. The explicit form of t(x) makes it possible to determine
exact solutions to the Requirements 14 posed in Part 1 of this project.

Figure S6. Graph of t(x) for = 0.1, A = 1.5, P = 15, and V = 6.

Part 3: What Do You Recommend?


Requirement 7
It is easy to design a machine that meets the given specications by using
the maximum allowed value of each parameter. Then the machine extracts 2 L
of fresh water from 8 L of brine in 18.4 h. The lter size can be reduced to an
area of 0.92 m2 and the machine will still satisfy the fresh water production
requirement. Table 1 gives examples of other results, from trial and error.
Table 1.
Various results.

Time (h) for 2 L


of fresh water

0.08
0.08
0.07

1.0
1.1
1.2

27
27
28

6
6
6

25.0
22.7
22.9

Students should discuss reasons for their choices of the parameters.

Getting the Salt Out

445

Title: Getting the Salt Out

Notes for the Instructor


Background on Reverse Osmosis
While reverse osmosis is sometimes referred to as a ltration process, many
researchers and practitioners prefer not to use this term due to the unique
aspects of this membrane separation. The mechanism of the separation is the
subject of some debate. The most widely accepted theory is that the process
proceeds by a solution-diffusion mechanism [Cussler 1997]. The membrane
does not have physical holes in it but rather acts as a distinct phase. Both
the water and the solute (salt in our case) dissolve in the membrane the same
as they might in an adjacent immiscible liquid phase. The components then
diffuse through the membrane and nally dissolve into the outlet liquid phase.
This is a rate-based separation where the water is puried because it diffuses
more easily, and therefore more quickly, through the membrane than the salt
does. The ux of water through the membrane is proportional to the pressure
above osmotic. The ow of solute is proportional to the concentration difference
across the membrane.
The example discussed in this project is a dead end batch conguration
of the separation process where all the water is being pushed through the
membrane. In practice, almost all reverse osmosis systems are designed in a
continuous cross-ow arrangement where the ow of the liquid is tangential to
the surface of the membrane. In this arrangement, there are two outlet streams,
a puried stream of the water that has passed through the membrane and a
concentrated stream of the material that has been retained by the membrane.
Either of these streams may be the product depending on whether the goal is
to produce puried water or a concentrated product.
Water purication is the largest use of reverse osmosis. However, this process is also used for concentrating various products, including orange juice
in the production of frozen concentrate and maple sap in the production of
maple syrup (where the approximately 3% solids sap must be raised to over
65% solids in the nal syrup). In the cross-ow conguration, the temporal
problem of increasing salt concentration explored in this paper becomes a spatial problem where the concentration is increasing as we move along the lter.
More background on this separation can be found in a short article by Eykamp
[1997] or in an article on the engineering project related to this problem [Moor
et al. 2004].
Vant Hoffs equation provides only an approximation of the osmotic pressure, and the approximation is best for low concentrations of solute. In fact,
the equation overpredicts the osmotic pressure by 35% in the range of concentrations used in this project. Furthermore, our model neglects the fact that
a small amount of salt migrates through the lter along with the water. This is

446

The UMAP Journal

24.4 (2003)

relatively inconsequential early in the extraction process. However, as water


is removed from the brine, the concentration of salt in the remaining solution
increases and, consequently, the difference between the applied pressure and
the osmotic pressure decreases. Eventually, while the rate at which water is
being extracted from the solution is decreasing, the rate of ow of salt through
the membrane actually goes up.

Notes on the Problem Solution


One can get a good idea of the nature of solutions of the desalination differential equation,

cV RT
dx
= A P
,
dt
V x
by examining its form. Initially, the rate of desalination is essentially directly
proportional to both and A. Furthermore, if the amount of fresh water that
has been removed from the salt water is small compared with the total volume
of salt water, then
P

cV RT
P cRT = P .
V x

Therefore, at the beginning of the desalination process, the rate at which fresh
water is being extracted is nearly constant. This is consistent with the solution
curves generated in the Sample Solution. As the value of x increases, V x
decreases and P cV RT /(V x) gets closer to zero. Therefore, as more and
more fresh water is extracted, the rate of extraction decreases.
The equilibrium solution of the differential equation, the constant value of x
for which dx/dt = 0, is of interest. Physically, equilibrium occurs when enough
fresh water has been removed from the solute so that the osmotic pressure
equals the applied pressure. Mathematically, equilibrium occurs when


cRT
,
x=V 1
P
and this value represents the maximum amount of fresh water that can be
extracted from a solution of given volume using a given applied pressure P .
The domain of the solution

cV RT
P x
1
x
ln 1 +
t(x) =
AP
P
cV RT V P

is
0x<V

cRT
1
P

There is a more rigorous way to determine appropriate parameter values


for this project. Start by making a nonoptimal choice for two of the parameter values, say A = 1.1 and V = 7. Now substitute these values in the

Getting the Salt Out

447

formula for t(x) and evaluate at x = 2. The resulting expression is a function


of two variables z(, P ). Graph z(, P ) and z = 1 together; the portion
of z(, P ) that lies below z = 1 contains permissible values for and P .
For example, Figure S7 shows the permissible values of 0.04 0.08 and
25 P 30 when A = 1.1 and V = 7. Therefore, the values (A, V, , P ) =
(1.15, 7.5, 0.075, 29) (0.05, 0.5, 0.005, 1) all result in an acceptable desalination
machine.

Figure S7. Determining permissible values for and P .

Pedagogical Advice
The instructor has a number of options concerning the administration of
this project. The project can be assigned all at once. As an alternative, when
we gave this project to our students, we spread the objectives out over a weeks
time. We spent part of a class period deriving the desalination differential
equation, to ensure that our students were familiar with the meaning of each
parameter. We then assigned Part 2 of the project, the explicit solution of the
differential equation, to be handed in before the laboratory session in which the
students worked on Part 1. We supplied the correct solution to the differential
equation within the lab itself. In this way, an incorrect solution to the differential
equation did not hamper any students analysis of the problem.
This project was developed jointly by engineering faculty and mathematics
faculty at Lafayette College. During the course of the semester, rst-year engineering students built and tested the desalination pump shown in Figure S8.

448

The UMAP Journal

24.4 (2003)

Figure S8. Desalination pump designed by rst-year engineering students.

We designed the mathematical analysis to give students an idea of the types


of important questions they could answer using basic differential equations.
In the semester-end evaluations, a number of the calculus students specically
mentioned that they appreciated seeing this application of the material. We note
that the parameter values we chose for this project were reasonably consistent
with the machine our students built, although the water permeability constant
is about 10 times actual value; the instructor should feel free to experiment with
other values as he or she chooses.
The graphs in the solution to this project were produced using Mathematica.
The slope eld package that generated the graphs is freely available; contact
the authors for a copy.

References
Cussler, E.L. 1997. Membranes. Chapter 17 in Diffusion: Mass Transfer in Fluid
Systems. 2nd ed. New York: Cambridge University Press.
Eykamp, W. 1997. Reverse osmosis and nanoltration. In Perrys Chemical
Engineers Handbook, 7th ed., edited by R.H. Perry, D.W. Green, and J.O.
Maloney, 22-4822-52. New York: McGraw-Hill.
Moor, S. Scott, Edmond P. Saliklis, Scott R. Hummel, and Yih-Choungh Yu.
2004. A press PO system: An interdisciplinary project for rst year engineering students. Chemical Engineering Education 37(1) (Winter 2003): 3844.
Nobel Foundation. 2000. The Nobel Prize in Chemistry 1901. http://www.
nobel.se/chemistry/laureates/1901/

Getting the Salt Out

449

About the Authors


Ethan Berkove is Assistant Professor of Mathematics at Lafayette College. Ethan studied mathematics and
Japanese as an undergraduate at the University of Michigan, then shifted over to mathematics for his Ph.D. at the
University of Wisconsin, Madison. His research eld is
algebraic topology, but his interests include applications
of mathematics and mathematical recreations.

Thomas Hill is Professor of Mathematics at Lafayette

College. Tom received his B.S from


North Carolina State
University and his Ph.D. from the University of Virginia.
He teaches a variety of mathematics courses, including
courses in mathematical modeling.

Scott Moor is Assistant Professor of Chemical Engineering at Lafayette College. He received B.S. and
M.S. degrees in Chemical Engineering from M.I.T. After
over a decade in industry, he returned to academia at the
University of California at Berkeley, where he received
a Ph.D. in Chemical Engineering and an M.A. in Statistics. He is a registered Professional Chemical Engineer
in the State of California. His research interests include
the development of materials for science, mathematics
and engineering education, applied statistics, and visualization of spray and particle system dynamics.

450

The UMAP Journal

24.4 (2003)

Vehicle Emissions

N T E R D I S C I P L I N A R Y

AUTHORS:
Lei Yu
(Transportation Studies)
yu_lx@tsu.edu
Texas Southern University
3100 Cleburne Avenue
Houston, TX 77004
Don Small
(Mathematics)
U.S. Military Academy
West Point, NY 10996
Lafayette College
Easton, PA 18042

CONTENTS
1. Introduction
2. Remote Emission-Sensing
Technique
3. Emission Data Collection
4. Data Conversion
5. Regression Analysis
6. Emission-Based Speed
Optimization
7. Real-World Applications
References
Sample Solution
Notes for the Instructor
About the Authors

I V E L Y

P P L I C A T I O N S

451

R O J E C T

Vehicle Emissions
MATHEMATICS CLASSIFICATIONS:
Algebra
DISCIPLINARY CLASSIFICATIONS:
Traffic Engineering
PREREQUISITE SKILLS:
1. Elementary algebra
2. Regression techniques
3. Elementary optimization

PHYSICAL CONCEPTS EXAMINED:


1. Concept of a mobile source for vehicle emissions
2. Optimal driving speeds to minimize emissions
3. Total vehicle emissions for a given speed profile
COMPUTING REQUIREMENT:
1. Spreadsheet and ability to use it
2. Calculator and ability to use it

The UMAP Journal 24 (4) 451-472. Copyright 2003 by COMAP, Inc. All rights
reserved. Permission to make digital or hard copies of part or all of this work for personal or classroom use is granted without fee provided that copies are not made or
distributed for profit or commercial advantage and that copies bear this notice.
Abstracting with credit is permitted, but copyrights for components of this work
owned by others than COMAP must be honored. To copy otherwise, to republish, to
post on servers, or to redistribute to lists requires prior permission from COMAP.

452

24.4 (2003)

Contents
1. Introduction
2. Remote Emission-Sensing Technique
3. Emission Data Collection
4. Data Conversion
5. Regression Analysis
6. Emission-Based Speed Optimization
7. Real-World Applications
References
Sample Solution
Notes for the Instructor
About the Authors

1.

Introduction

Cars, trucks, motorcycles, and buses emit signicant quantities of carbon


monoxide (CO), hydrocarbons (HC), nitrogen oxides (NOx), and ne particles
(PM). These chemical compounds play dominant roles in air pollution problems. In the densely populated Northeast, where the air pollution problem is
especially severe, the Environmental Protection Agency (EPA) has projected
that highway vehicles will account for approximately 38% of the total NOx inventory and 22% of the total HC inventory in 2005, in spite of the tighter motor
vehicle standards in the 1990 Clean Air Act Amendment (CAAA) (Figure 1).

Figure 1. Trafc.

Vehicle emissions, especially CO, NOx, and HC, are harmful to the health
of humans. CO is a tasteless, odorless, and colorless gas produced through
the incomplete combustion of carbon-based fuels. CO enters the bloodstream
through the lungs and reduces the delivery of oxygen to the bodys organs
and tissues. The health threat from exposure to CO is most serious for those

Vehicle Emissions

453

who suffer from cardiovascular disease, particularly those with angina or peripheral vascular disease. Exposure to elevated CO levels is associated with
impairment of visual perception, work capacity, manual dexterity, learning
ability and performance of complex tasks. NOx emissions also produce a wide
variety of health and welfare effects. Nitrogen dioxide can irritate the lungs and
lower resistance to respiratory infection (such as inuenza). NOx emissions are
important in acid rain, affecting both terrestrial and aquatic ecosystems. Atmospheric deposition of nitrogen leads to excess nutrient enrichment problems
(eutrophication). Finally, HC, in combination with NOx in the presence of
heat and sunlight, will form ozone. Ozone forms readily in the atmosphere,
usually during hot summer weather. Ground-level ozone is the prime ingredient of smog, the pollution that blankets many areas during the summer.
Short-term exposures (13 hours) to high ambient ozone concentrations have
been linked to increased hospital admissions and emergency room visits for
respiratory problems.
With the rapid increase of the number of motor vehicles along with worldwide economic development, the vehicle emission problem has worsened.
Many governments have started to implement certain vehicle emission-control
strategies. Steady progress in reducing certain air pollution problems is occurring in the US, Europe and Japan. Globally, the use of advanced pollutioncontrol technology, especially catalysts has been spreading, as has the use of
unleaded gasoline. However, the continued economic growth in the world requires the use of strategies that do not impose negative effect on the economic
development, in other words, do not restrict vehicle ownership. Examples
of such strategies that have already been implemented in many cities in the
world include advanced trafc-control techniques, integrated transportation
planning process, and Intelligent Transportation Systems (ITS) technologies.
Past research has shown that driving patterns greatly inuence the amount
of vehicle emissions. Frequent acceleration and deceleration tend to generate
more emissions than smoother driving. An effective trafc signal timing plan
can smooth trafc ow in a manner that reduces the emissions. In addition, well
planned transportation projects or activities can change the driving patterns of
vehicles in the city at a more macroscopic level.

2.

Remote Emission-Sensing Technique

The key to success in developing an emission-control strategy is obtaining


accurate quantication of vehicle emissions under various trafc and environmental scenarios. The remote emission-sensing device that is used in this
project is called SMOG DOG. It performs environmental monitoring to measure
automotive emissions. The SMOG DOG can simultaneously measure emission
concentrations of CO, HC, NOx, and CO2 in the dispersing exhaust cloud of
vehicles. A special feature of the SMOG DOG is its capability in detecting a
vehicles instantaneous speed and acceleration rate.

454

24.4 (2003)

Figure 2. SMOG DOG in action.

3.

Emission Data Collection

SMOG DOG was used to collect vehicle emission data from a freeway onramp in Houston, TX (Figure 2). The operation of SMOG DOG requires that
the vehicle be in motion. Hence, the emission data for the idling mode are
not collected. Instead, vehicle idling emissions will be calculated based on the
regression models that are developed. Table 1 illustrates the emission data on
CO and HC concentrations that were collected.
Problem: Develop models relating vehicle speed and vehicle exhaust emissions
of CO and HC. Formulate speed recommendations based on your model and
the given data. Use of spreadsheet technology is strongly recommended. Input
the data in Table 1 into a spreadsheet (4 columns). You will add four more
columns in Requirement 2 and two more in Requirement 4.

Requirement 1
Use the data in Table 1 to
plot the CO concentration (%) vs. speed data, and
plot the HC concentration (%) vs. speed data.

Vehicle Emissions

455

Table 1.
CO and HC concentration data collected by SMOG DOG in Houston.
Data #
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30

4.

Speed (mph)

CO%

HC%

9.98
10.97
15.40
15.69
16.44
17.11
18.66
19.04
20.95
21.52
23.41
24.82
25.15
26.35
27.02
27.62
29.90
30.04
30.67
31.99
34.97
35.10
35.94
36.98
37.69
37.90
38.74
40.58
41.63
42.12

0.27
0.22
0.30
0.03
0.21
0.89
0.19
0.79
0.08
0.07
1.61
0.77
0.76
1.91
0.40
0.06
0.36
0.18
0.58
0.12
1.02
0.53
1.11
0.29
1.67
0.62
0.27
0.09
0.30
0.54

0.000027
0.000022
0.00003
0.000003
0.000021
0.000089
0.000019
0.000079
0.000008
0.000007
0.000161
0.000077
0.000076
0.000191
0.00004
0.000006
0.000036
0.000018
0.000058
0.000012
0.000102
0.000053
0.000111
0.000029
0.000167
0.000062
0.000027
0.000009
0.00003
0.000054

Data #
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Speed (mph)

CO%

HC%

43.22
44.20
45.05
46.59
47.06
47.51
47.86
48.92
49.78
49.81
50.50
50.52
51.48
52.18
52.87
53.25
53.62
54.32
55.19
56.07
56.33
56.56
58.29
58.75
59.44
61.40
62.00
63.65
67.11
76.68

0.32
0.29
0.39
0.79
0.73
0.54
0.07
0.33
0.43
0.55
0.72
0.21
0.18
0.61
0.92
0.19
0.44
0.22
0.34
0.49
0.15
0.76
0.12
0.12
0.35
0.83
0.21
0.35
0.35
0.47

0.000032
0.000029
0.000039
0.000079
0.000073
0.000054
0.000007
0.000033
0.000043
0.000055
0.000072
0.000021
0.000018
0.000061
0.000092
0.000019
0.000044
0.000022
0.000034
0.000049
0.000015
0.000076
0.000012
0.000012
0.000035
0.000083
0.000021
0.000035
0.000035
0.000047

Data Conversion

To develop models relating vehicle speed to emission quantities, we introduce two emission quantities: emission factor (g/mi) and emission rate (g/s).
The emission factor is the number of grams emitted when the vehicle
travels one mile.
The emission rate is the total emission in grams that a vehicle emits
per second.
We dene the following variables:
CO%: CO emission concentration (%);
HC%: HC emission concentration (%);
COs: CO emission rate in grams per second (g/s);

456

24.4 (2003)

HCs: HC emission rate in grams per second (g/s);


COm: CO emission factor in grams per mile (g/mi);
HCm: HC emission factor in grams per mile (g/mi);
u: a vehicles instantaneous speed in miles per hour (mph);
uoptimal : optimal speed (mph) that minimizes emissions;
b0 , b1 , c0 , c1 : constant values (regression model coefcients).
Conversion of emission concentrations to emission rates is problematic.
Many scientists and engineers use the following linear correlation relationships,
developed for the emission concentrations and emissions factors by the South
Coast Air Quality Management District (SCAQMD).
Conversion from emission concentrations to emission factors:
COm (g/mi) = 11.1 CO% + 21.3,

(1)

HCm (g/mi) = 63.3 HC% + 1.7.

(2)

Conversion from emission concentrations to emission rates:


COm (g/mi) u mph
,
3600
HCm (g/mi) u mph
HCs (g/s) =
.
3600

COs (g/s) =

(3)
(4)

Requirement 2
Use (1)(4) to convert the CO/HC concentrations in Table 1 to CO/HC
emission factors and CO/HC emission rates. Expand your spreadsheet table
to include the 8 columns: Data #, Speed, CO%, HC%, COm, HCm, COs, and
HCs.

Requirement 3
Draw four graphs to show the relationships between speed and CO/HC
emission factors as well as CO/HC emission rates using the data prepared in
Requirement 2. Based on your visual analysis of these graphs, which relationships display a better regularity of shapes and trends; emission factors or
emission rates?

Vehicle Emissions

5.

457

Regression Analysis

In many engineering and scientic applications, relationships between variables are established by rst collecting data from experimental studies in either
the laboratory or the eld. This data is then plotted and relationship discerned.
In this project, the two variables in the relationship represent the speed and
the CO/HC emission rates. Because of potential errors in measurement, the
data shown in the graphs drawn in Requirement 3 may not fall precisely on a
smooth curve. For this reason, the task of the analysis becomes threefold:
to hypothesize the mathematical form of the relationship between the two
variables (model postulation);
to estimate the parameters of the model based on the collected eld data
(model calibration); and
to determine how well the calibrated relationship explains the observed data
(goodness of t).
This analysis process is called regression.
Scientic research on vehicle emissions has shown that the best mathematical function to model the relationship between the speed and the CO or HC
emission rate is the natural logarithm, expressed in the following forms:
ln(COs) = b0 + b1 u,
ln(HCs) = c0 + c1 u.

(5)
(6)

The speed u is the independent or explanatory variable. The emission rates, COs
or HCs, are called the dependent or explained variables. The constant coefcients
c0 and c1 are the model parameters. Data from Requirement 2 can be used to
plot ln(COs) vs. speed and ln(HCs) vs. speed. Calibrating the model means
determining the unknown values of the parameters in (5) or (6) to obtain the
best t to these plots. The goodness of t is measured by the coefcient of
correlation, r. Its value can range from 1 to +1. If r is near +1, there exists a
high positive correlation; if it is near 1, there exists a high negative correlation;
and if it is around zero, there exists no correlation between the independent
and the dependent variables.

Requirement 4
Use the regression technique and the data prepared in Requirement 2 to
calibrate (5) and (6). The data for COs and HCs prepared in Requirement 2
need to be transformed by the natural logarithm (ln) before a simple linear
regression can be carried out. Hint: Add columns for ln(C0s) and ln(HCs) to
your spreadsheet table constructed in Requirement 2. Then plot the ln(C0s)
vs. speed and form the linear regression. Repeat for ln(HCs) vs. speed.

458

24.4 (2003)

Requirement 5
Calculate the coefcient correlations for the regressions in Requirement 4.
Are the models acceptable? Why? Draw the line plot for each of the regression
equations generated.

6.

Emission-Based Speed Optimization

Emission rates (g/s) can be transformed into emission factors (g/mi) by


solving (3) for COm and (4) for HCm. The new equations are:
COs (g/s) 3600 (s/h)
,
u (mi/h)
HCs (g/s) 3600 (s/h)
HCm (g/mi) =
.
u (mi/h)
COm (g/mi) =

(7)
(8)

Equations (3) and (4) for the emission rates are in the form y = mx, where the
slopes are COm/3600 and HCm/3600. Because these slopes are positive, the
emission rates are monotonically increasing functions of speed; hence, there
are no minimum values for a moving vehicle. The situation is less clear for the
emission factors described by (7) and (8). These equations can be transformed
into explicit functions of speed by solving for COs and HCs in the results of
Requirement 4 and then substituting these results into (7) and (8). Doing this
yields
3600eb0 +b1 u
,
COs (g/s) =
u
3600ec0 +c1 u
HCm (g/s) =
.
u

(9)
(10)

A graphical analysis can now be made to determine what speeds will minimize the emission factors. Knowledge of these speeds is critical information
for developing an advanced trafc management scheme to reduce emission
amounts.

Requirement 6
Draw a graph with two curves, one for the relationship between the speed
and the CO emission rate, and the other for the relationship between the speed
and the CO emission factor. Draw a second graph with two curves, one for the
relationship between the speed and the HC emission rate, and the other for the
relationship between the speed and the HC emission factor. Let the speed be
measured starting from 0 to 70 mph in increments of 5 mph. Then calculate
the emission rates by solving for COs and HCs in the equations generated in
Requirement 4, and the emission factors using (9) and (10). Based on the visual
analysis of the graphs, is there a speed that can minimize the emission factor?

Vehicle Emissions

459

Requirement 7
Approximate the speeds corresponding to the minimum points on the emission factor plots. What speed do you recommend for minimizing both the CO
and HC emission factors? Hint: On a graphing calculator, use the trace feature
to trace a point along the curve to the minimum position or use the table feature
to determine the coordinates of the minimum point. For a spreadsheet, iterate
the function values to determine the minimum.

7.

Real-World Applications

If all drivers traveled at the optimal speed, the total CO or HC emissions


would be the minimum. On the other hand, if vehicles are driven at varying
speeds, the resulting emissions will be higher. To analyze such a pattern, we assume two different speed proles experienced by two different vehicles. Both
vehicles are driven for 30 s, but the rst vehicle travels at a constant speed of
35 mph while the speed of the second vehicle varies between 29 and 42 mph (see
Table 2). After 30 s, both vehicles have driven the same distance, 1540 ft. Although both vehicles have been driven the same distance during the same time
period, it is expected that the resulting vehicle emissions would be different.
Table 2 illustrates the details of these two different speed proles.

Requirement 8
Based on Table 2, plot two graphs, one for the constant speed prole, and
one for the varying speed prole. For each speed prole, there should be
two curves, one for the relationship between time and speed, and one for the
relationship between time and the distance. Use different y-axes for speed and
distance.

Requirement 9
Use the regression equations derived from Requirement 4 to calculate the
CO and HC emissions at each second for the two different speed proles in
Table 2. Make your calculation in a table. Then, calculate the total CO and HC
emissions for the entire travel of the two speed proles. What can you conclude
from your calculations?

Requirement 10
Assume that an accident happened on a location of a freeway segment. A
police ofcer came to the scene to process the necessary paperwork for this
accident and towing vehicles came to remove all vehicles involved. It took a

460

24.4 (2003)
Table 2.
Speed proles driven by two vehicles.
Time (Sec)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30

Constant Speed Profile


Speed (mph)
Distance (ft)
35
51
35
103
35
154
35
205
35
257
35
308
35
359
35
411
35
462
35
513
35
565
35
616
35
667
35
719
35
770
35
821
35
873
35
924
35
975
35
1027
35
1078
35
1129
35
1181
35
1232
35
1283
35
1335
35
1386
35
1437
35
1489
35
1540

Varying Speed Profile


Time (Sec)
Speed (mph)
Distance (ft)
1
42
62
2
41
122
3
40
180
4
39
238
5
38
293
6
38
349
7
37
403
8
36
456
9
35
507
10
34
557
11
33
606
12
32
653
13
31
698
14
30
742
15
30
786
16
30
830
17
29
873
18
29
915
19
30
959
20
31
1005
21
32
1052
22
33
1100
23
34
1150
24
35
1201
25
36
1254
26
37
1308
27
38
1364
28
39
1421
29
40
1480
30
41
1540

total of 40 min from the start of the accident to the time when the accident scene
is completely cleared. During this period, a total of 200 vehicles have to stop.
Assume the average wait is 30 min per vehicle. Calculate the total additional
CO and HC emissions that were generated by these 200 vehicles due to the
trafc accident. What can you conclude from your calculation? Hint: Derive
the idling emissions rates by setting the speed equal to zero in the results of
Requirement 4.

References
European Conference of Ministers of Transport (ECMT). 291. Vehicle Emission
Reductions. Paris, France: OECD Publications Services (2, rue Andre Pascal,
75775 Paris Cedex 16, France).
Sorbe, N. 1995. Hughes Employee Vehicle Exhaust Remote Sensing and Emissions Evaluation Project. Report prepared for the Mobile Source Air Pollution Reduction Review Committee (MSRC) under the AB2766 Program. El
Segundo, CA: Hughes Environmental Systems, Inc.

Vehicle Emissions

461

Yu, L. 1998a. Collection and Evaluation of Modal Trafc Data for Determination
of Vehicle Emission Rates under Certain Driving Conditions. Research
Reports 1485-1, 1485-2, and 1485-3F for TxDOT Project No. 0-1485, Center
for Transportation Training and Research. Houston, TX: Texas Southern
University.
. 1998b. Remote vehicle exhaust emission-sensing for trafc simulation and optimization models. Journal of Transportation Research Part D:
Transport and Environment 3: 337347.

462

24.4 (2003)

Title: Vehicle Emissions

Sample Solution
Requirement 1: The graphs between the speed and CO and HC concentrations
are shown in Figures S1 and S2.
CO% Versus Speed
2.5

CO Concentration (%)

1.5

0.5

0
0

10

20

30

40

50

60

70

80

90

Speed (mph)

Figure S1. CO concentration vs. speed.

Based on visual analysis of these graphs, clearly the speed is not correlated
with the CO or HC concentrations. The data are scattered and do not show any
regularity of shapes or trends.
Requirement 2: The generated table is shown in Table S1.
Requirement 3: See Figures S3S6. A visual analysis of the graphs suggests
that the CO/HC emission rates have greater regularity of shape and trend with
the speed than do the CO/HC emission factors.
Requirement 4: The resulting regression equations are:
ln(COs) = 2.46215 + 0.028383u,

or

COs = e2.46215+0.028383u ,

ln(HCs) = 4.91624 + 0.030929u,

or

COs = e4.91624+0.030929u .

Vehicle Emissions

463

HC% Versus Speed


0.04

0.035

HC Concentration (%)

0.03

0.025

0.02

0.015

0.01

0.005

0
0

10

20

30

40

50

60

70

80

90

80

90

Speed (mph)

Figure S2. HC concentration vs. speed.

CO (g/mi) Versus Speed


45.0000

40.0000

35.0000

CO Emissions (g/mi)

30.0000

25.0000

20.0000

15.0000

10.0000

5.0000

0.0000
0

10

20

30

40

50

60

Speed (mph)

Figure S3. CO emission factor vs. speed.

70

464

24.4 (2003)
Table S1.
Speed proles driven by two vehicles.
Data #

Speed

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

CO%

HC%

COm(g/mi) HCm(g/mi) COs(g/sec)

9.98
10.97

0.27
0.22

0.0096 24.2970
0.0058 23.7420

15.40
15.69

0.30
0.03

0.0038 24.6300
0.0031 21.6330

16.44
17.11
18.66

0.21
0.89
0.19

0.0161 23.6310
0.0031 31.1790
0.0055 23.4090

19.04
20.95

0.79
0.08

0.0241 30.0690
0.0065 22.1880

21.52
23.41

0.07
1.61

0.0007 22.0770
0.0063 39.1710

24.82
25.15

0.77
0.76

0.0187 29.8470
0.0257 29.7360

26.35
27.02

1.91
0.40

0.0370 42.5010
0.0206 25.7400

27.62
29.90

0.06
0.36

0.0050 21.9660
0.0256 25.2960

30.04
30.67
31.99

0.18
0.58
0.12

0.0041 23.2980
0.0036 27.7380
0.0161 22.6320

34.97
35.10

1.02
0.53

0.0013 32.6220
0.0084 27.1830

35.94
36.98

1.11
0.29

0.0122 33.6210
0.0061 24.5190

37.69
37.90

1.67
0.62

0.0350 39.8370
0.0206 28.1820

38.74
40.58

0.27
0.09

0.0152 24.2970
0.0037 22.2990

41.63
42.12

0.30
0.54

0.0063 24.6300
0.0133 27.2940

43.22
44.20
45.05

0.32
0.29
0.39

0.0147 24.8520
0.0141 24.5190
0.0247 25.6290

46.59
47.06

0.79
0.73

0.0239 30.0690
0.0079 29.4030

47.51
47.86

0.54
0.07

0.0171 27.2940
0.0071 22.0770

48.92
49.78

0.33
0.43

0.0248 24.9630
0.0345 26.0730

49.81
50.50

0.55
0.72

0.0150 27.4050
0.0138 29.2920

50.52
51.48

0.21
0.18

0.0173 23.6310
0.0056 23.2980

52.18
52.87
53.25

0.61
0.92
0.19

0.0352 28.0710
0.0186 31.5120
0.0282 23.4090

53.62
54.32

0.44
0.22

0.0013 26.1840
0.0329 23.7420

55.19
56.07

0.34
0.49

0.0211 25.0740
0.0054 26.7390

56.33
56.56

0.15
0.76

0.0039 22.9650
0.0351 29.7360

58.29
58.75

0.12
0.12

0.0169 22.6320
0.0103 22.6320

59.44
61.40

0.35
0.83

0.0038 25.1850
0.0068 30.5130

62.00
63.65
67.11

0.21
0.35
0.35

0.0084 23.6310
0.0068 25.1850
0.0089 25.1850

76.68

0.47

0.0073 26.5170

2.3077
2.0671
1.9405
1.8962
2.7191
1.8962
2.0482
3.2255
2.1115
1.7443
2.0988
2.8837
3.3268
4.0421
3.0040
2.0165
3.3205
1.9595
1.9279
2.7191
1.7823
2.2317
2.4723
2.0861
3.9155
3.0040
2.6622
1.9342
2.0988
2.5419
2.6305
2.5925
3.2635
3.2129
2.2001
2.7824
2.1494
3.2698
3.8839
2.6495
2.5735
2.7951
2.0545
3.9282
2.8774
3.4851
1.7823
3.7826
3.0356
2.0418
1.9469
3.9218
2.7698
2.3520
1.9405
2.1304
2.2317
2.1304
2.2634
2.1621

0.0674
0.0723
0.1054
0.0943
0.1079
0.1482
0.1213
0.1590
0.1291
0.1319
0.2547
0.2057
0.2077
0.3110
0.1932
0.1685
0.2101
0.1944
0.2363
0.2011
0.3169
0.2650
0.3356
0.2518
0.4171
0.2967
0.2615
0.2514
0.2848
0.3193
0.2983
0.3010
0.3207
0.3891
0.3844
0.3602
0.2935
0.3392
0.3605
0.3791
0.4109
0.3316
0.3332
0.4068
0.4627
0.3462
0.3900
0.3582
0.3844
0.4165
0.3593
0.4671
0.3664
0.3693
0.4158
0.5204
0.4070
0.4452
0.4695
0.5648

HCs(g/sec)

0.0064
0.0063
0.0083
0.0083
0.0124
0.0090
0.0106
0.0171
0.0123
0.0104
0.0136
0.0199
0.0232
0.0296
0.0225
0.0155
0.0276
0.0163
0.0164
0.0242
0.0173
0.0218
0.0247
0.0214
0.0410
0.0316
0.0286
0.0218
0.0243
0.0297
0.0316
0.0318
0.0408
0.0416
0.0288
0.0367
0.0286
0.0444
0.0537
0.0367
0.0361
0.0392
0.0294
0.0569
0.0423
0.0515
0.0265
0.0571
0.0465
0.0318
0.0305
0.0616
0.0448
0.0384
0.0320
0.0363
0.0384
0.0377
0.0422
0.0460

Vehicle Emissions

465

HC (g/mi) Versus Speed


4.5000

4.0000

3.5000

HC Emissions (g/mi)

3.0000

2.5000

2.0000

1.5000

1.0000

0.5000

0.0000
0

10

20

30

40

50

60

70

80

90

70

80

90

Speed (mph)

Figure S4. HC emission factor vs. speed.

CO (g/sec) Versus Speed


0.6000

0.5000

CO Emissions (g/sec)

0.4000

0.3000

0.2000

0.1000

0.0000
0

10

20

30

40

50

60

Speed (mph)

Figure S5. CO emission rate vs. speed.

466

24.4 (2003)

HC (g/sec) Versus Speed


0.0700

0.0600

0.0400

0.0300

0.0200

0.0100

0.0000
0

10

20

30

40

50

60

70

80

90

Speed (mph)

Figure S6. HC emission rate vs. speed.

Requirement 5:
r(COs) = 0.9089,
r(HCs) = 0.8644.
Both correlation coefcients are close to one, so both regression equations
have acceptable goodness of t. Since r(COs) is slightly higher than r(HCs),
the regression equation for COs is better than the one for HCs.
Speed Line Plot

ln(CO)

HC Emissions (g/sec)

0.0500

0
-2
-4

50

100

LN(CO)
Predicted LN(CO)

Speed

Figure S7. Line plot for the speed and the CO emission factor.

Requirement 6: From Figures S9S10, there clearly exists a speed value that
minimizes either the CO emission factor or the HC emission factor. On the
other hand, the emission rates are monotonically increasing functions of the
speed.
Requirement 7:

uoptimal for CO = 35 mph;

uoptimal for HC = 32 mph.

Vehicle Emissions

ln(HC)

Speed Line Plot

0
-5 0

50

LN(HC)

100

Predicted LN(HC)

-10

Speed

Figure S8. Line plot for the speed and the HC emission factor.

CO Emissions Versus Speed


0.70

70

CO (g/mi)

0.60

60

CO (g/sec)

0.50

50

0.40

40

0.30

30

CO (g/sec)

CO (g/mi)

80

0.20

20

0.10

10
0

0.00
0

20

40

60

80

Speed (mph)

Figure S9. Relationships between the CO emission rates/factors and the speed.

HC Emissions Versus Speed


7

0.07

0.06
HC (g/mi)

0.05

HC (g/sec)

0.04

0.03

0.02

0.01

HC (g/sec)

HC (g/mi)

0.00
0

20

40

60

80

Speed (mph)

Figure S10. Relationships between the HC emission rates/factors and the speed.

467

468

24.4 (2003)

Requirement 8: See Figures S11S12.

1800

40

1600

35

1400

30

1200

Distance (ft)
Speed (mph)

1000

25
20

800
15

600
400

10

200

Speed (mi/hr)

Accumulative Distance (ft)

Speed Profile 1 - Constant Speed

0
0

10

15

20

25

30

Time (sec)

Figure S11. Illustration of constant speed prole.

1800

45

1600

40

1400

35

1200

30

1000

25

Distance (ft)
Speed (mph)

800

20

600

15

400

10

200

0
0

10

15

20

25

Time (sec)

Figure S12. Illustration of varying speed prole.

30

Speed (mi/hr)

Accumulative Distance (ft)

Speed Profile 2 - Varying Speed

Vehicle Emissions

469

Requirement 9: See Table S2.


Table S2.
Second-by-second calculations.
Time
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30

Constant Speed Prole


Speed 1
CO 1
HC 1
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35
35

TOTAL EMISSIONS

0.2302
0.2302
0.2302
0.2302
0.2302
0.2302
0.2302
0.2302
0.2302
0.2302
0.2302
0.2302
0.2302
0.2302
0.2302
0.2302
0.2302
0.2302
0.2302
0.2302
0.2302
0.2302
0.2302
0.2302
0.2302
0.2302
0.2302
0.2302
0.2302
0.2302

0.02163
0.02163
0.02163
0.02163
0.02163
0.02163
0.02163
0.02163
0.02163
0.02163
0.02163
0.02163
0.02163
0.02163
0.02163
0.02163
0.02163
0.02163
0.02163
0.02163
0.02163
0.02163
0.02163
0.02163
0.02163
0.02163
0.02163
0.02163
0.02163
0.02163

6.9064

0.6489

Varying Speed Prole


Speed 2
CO 2
HC 2
42
41
40
39
38
38
37
37
37
36
35
34
33
32
31
30
29
29
30
31
32
33
34
35
36
37
38
39
40
41

0.2808
0.2730
0.2653
0.2579
0.2507
0.2507
0.2437
0.2437
0.2437
0.2368
0.2302
0.2238
0.2175
0.2114
0.2055
0.1998
0.1942
0.1942
0.1998
0.2055
0.2114
0.2175
0.2238
0.2302
0.2368
0.2437
0.2507
0.2579
0.2653
0.2730

0.02686
0.02604
0.02525
0.02448
0.02373
0.02373
0.02301
0.02301
0.02301
0.02231
0.02163
0.02097
0.02033
0.01971
0.01911
0.01853
0.01797
0.01797
0.01853
0.01911
0.01971
0.02033
0.02097
0.02163
0.02231
0.02301
0.02373
0.02448
0.02525
0.02604

7.0382

0.66273

For the CO emissions, the varying speed prole generates a total of 7.04 g
comparing with the 6.91 g for the constant speed prole (2% higher). Similarly
for the HC emissions, the varying speed prole generates a total of 0.663 g comparing with the 0.649 g of emissions for the constant speed prole (2% higher).
Since the optimal speed is 35 mph for CO and 32 mph for HC, the constant
speed prole in this requirement should result in the minimum CO emissions
and close to minimum HC emissions. Although the 2% higher emissions does
not seem like a signicant amount, this is just for 30 s for a single vehicle. You
can imagine how much additional emissions would be generated if all vehicles
in the city drive at varying speeds of a signicant level. On the other hand, the
trafc managers should try to implement trafc management strategies that
would encourage or force drivers to drive at desired and constant speeds.

470

24.4 (2003)

Requirement 10: Set speed to 0 in the regression equations in Requirement 4


to derive the idling emission rate in units of grams per second:
COs when idling = e2.46215+0.028383u = 0.08525 g/s,
HCs when idling = e4.91624+0.030929u = 0.00733 g/s.
Therefore, during the 30 min delay, a vehicle will emit the following total CO
and HC emissions:
CO idling emissions for one vehicle for 30 min = 0.08525 30 60=153.45 g,
HC idling emissions for one vehicle for 30 min = 0.00733 30 60= 13.194 g.
Therefore, the total CO emissions for 200 vehicles = 153.45 200 = 30,690 g and
the total HC emissions for 200 vehicles = 13.194 200 = 2,639 g.
From the above calculation, we can see that one trafc accident can result
in considerable additional CO and HC emissions. Considerably more emissions will be produced if many trafc accidents happen in the city. Therefore,
reducing the number of trafc accidents can reduce the vehicle emissions.

Vehicle Emissions

471

Title: Vehicle Emissions

Notes for the Instructor


This project is designed for students to practice algebraic calculations, data
analysis, and optimization using real-world vehicle emission data. The vehicle
emission issue is becoming more and more important because of the stricter
regulations imposed by the U.S. Environmental Protection Agency (EPA). Vehicle emissions are a big source of air pollution and are related to many of our
health problems.
The key to the successful completion of this project is a clear understanding
of the sequence of the data, the denitions of the variables, the relationships
between variables, as well as the algebraic, data analysis, and optimization
concepts. Any additional information related to vehicle emission-control can
be introduced in the class. The instructor can introduce various emission data
collection techniques. The instructor can also introduce more complicated modeling and data analysis techniques.
The simplest regression equation is used in this project, which although
nonlinear, includes only one variable. The instructor can encourage students
to test various regression formulas and different combinations of variables. For
instance, other regression models to consider include
COs = b0 + b1 u + b2 u2 ,
HCs = c0 + c1 u + c2 u2 .
Requirements 49 can be repeated with these other regression formulas.
The instructor can also encourage students to search literature on other
existing vehicle emission models. Students can compare the calculation results
between the models developed in this project and other models.

472

24.4 (2003)

About the Authors


Lei Yu is Professor and Chairperson of the Transportation Studies Department at Texas Southern University. He is also a Changjiang Scholar of Beijing Jiaotong University. He received a bachelors degree in
Transportation Engineering from Northern Jiaotong University, Beijing, China and a masters degree in Production and Systems Engineering from Nagoya Institute of Technology, Nagoya, Japan. His Ph.D. degree
in Civil Engineering was awarded by Queens University in Kingston, Ontario, Canada. His research interests involve highway trafc operations and modeling,
urban transportation planning, the ITS-related technologies and applications,
and vehicle exhaust emission modeling. Since 1994, Yu has been the Principal Investigator (PI) of more than 25 research projects that were sponsored by
Texas Department of Transportation (TxDOT), Federal Highway Administration (FHWA), Federal Transit Administration (FTA), Southwest Region University Transportation Center (SWUTC) program, National Institute of Standards
and Technology (NIST), and Houston Advanced Research Center (HARC); the
total funding amount is over $2 million. Dr. Yu has published more than 50
research papers in the Journal of Transportation Research, Transportation Research
Records, and the Journal of Transportation Engineering, as well as in various peerreviewed conference proceedings. Dr. Yu is an active member of the Institute of
Transportation Engineers (ITE), the American Society of Civil Engineers (ASCE)
and the Transportation Research Board (TRB). He is a registered engineer in
the State of Texas. He also belongs to numerous committees, councils, and task
forces in regional, state, national, and international organizations.
Don Small graduated from Middlebury College and
received his Ph.D. degree in mathematics from the University of Connecticut. He taught mathematics at Colby
College for 23 years before joining the Mathematics Department at the U.S. Military Academy in 1991. He is active
in the calculus and college algebra reform movements
developing curricula, authoring texts, and leading faculty
development workshops. Active in the Mathematical Association of America (MAA), Don served a term as Chairman and two terms
as Governor of the Northeast Section and has been a long-term member of
the MAAs CRAFTY committee. His interests are in developing curriculum
that focuses on student growth while meeting the needs of partner disciplines,
society, and the workplace. He is also a member of AMATYC and AMS.

Reviews

473

Reviews
King, A.C., J. Billingham, and S.R. Otto. 2003. Differential Equations: Linear,
Nonlinear, Ordinary, Partial. New York: Cambridge University Press; 554 pp,
$100. ISBN 0521816580.
The American tradition in mathematics is rather different from the British.
An uncharitable characterization of the former would be that American mathematicians of the nineteenth century were largely hacks. J.J. Sylvester at Johns
Hopkins was an anomaly, and represented genuine scholarly excellence in the
European tradition. He was, after all, British. It is widely thought that G. D.
Birkhoff brought the rst serious attention to American mathematics (just because he proved theorems that were recognized in Europe). He was followed
by Lefschetz, Veblen, Bochner, Alexander, and many other powerful and inuential American scholars who established the pre-eminence of the U.S. in
modern mathematics.
Of course the history of British mathematics goes back much further. Isaac
Barrow was the rst Professor of Mathematics at Cambridge. He was followed by Isaac Newton, who made quite a splash. Newtons departure, as
we all know, left a considerable vacuumit can be said that British mathematics imploded for more than a century. But an important tradition sprang
up in the nineteenth century, and that is the British paradigm for concrete applied mathematics. Certainly one of the rst avatars of this point of view was
George Green. In his short but fruitful career, he invented the Greens function, Greens theorem, the idea of potential function, what is now known as
the Dirichlet principle, and the idea of the tensor. He had tremendous inuence on Cayley, Maxwell, and many other applied mathematicians in the late
nineteenth century. In spite of Hardy and Littlewoods wholesale rejection of
applications or uses of mathematics, the strong British applied tradition lives
on in practitioners like Mary Cartwright (recently deceased) and Elliott Lieb
(still very active and productive at Princeton).
The book under review is the product of the British practice in applied
mathematics. For this we can be grateful, for it represents a breath of fresh
air in the differential equations textbook market in this country. There are
not a great many undergraduate texts in partial differential equations in the
U.S., and precious few that have any real depth and substance. There are
fewer still rmly grounded in applicationsthe very wellspring of this part of
c
The UMAP Journal 24 (4) (2003) 473478. Copyright
2003 by COMAP, Inc. All rights reserved.
Permission to make digital or hard copies of part or all of this work for personal or classroom use
is granted without fee provided that copies are not made or distributed for prot or commercial
advantage and that copies bear this notice. Abstracting with credit is permitted, but copyrights
for components of this work owned by others than COMAP must be honored. To copy otherwise,
to republish, to post on servers, or to redistribute to lists requires prior permission from COMAP.

474

The UMAP Journal 24.4 (2003)

mathematics. This book, arising as it does from the British history described
above, is a valuable contribution and will be a pleasure to have as a resource
for our undergraduate (and even our beginning graduate) curriculum.
The authors have made a number of very smart choices in designing their
book. First, they assume that the reader has a little background in the basic
techniques of linear ordinary differential equations. That way they can get off
to a ying start on their subject proper.1 As they point out in their Preface, there
are two types of differential equations books on the market:
those that are far too theoretical, and
those that are too much plug and chug (offering explicit, concrete solution
techniques to a limited collection of rather dry old problems).
The authors profess to steer a clear course between these two markers, they let
the applications be their guide, and I believe that they succeed admirably.
The title of this book is already indicative of the direct, evenhanded approach
that this team of authors brings to their task. There is no nonsense, little ash,
but instead a refreshingly weighty dose of substance. In addition to studying
topics that you would expect (such as Greens functions, boundary value problems, Sturm-Liouville theory, Fourier series, Laplace transforms, phase-plane
analysis, and existence/uniqueness), they also delight us with detailed and
incisive treatments of Legendre functions, Bessel functions (can you detect the
British tradition, e.g., Watson?), Lie groups, asymptotic methods, time-optimal
control, chaotic systems, and bifurcations.
The book is chock full of really meaty applicationssome of which you
can present in class and some of which require further study and could be the
subject for a term paper or team project. These include van Dykes matching
principle, boundary layer potentials, thermal ignition, the van der Pol oscillator,
Kuzmaks method for slowly damped nonlinear oscillations, reaction-diffusion
processes, the WKB approximation, the Helmholtz equation, time-optimal trajectories, mechanical oscillators, and on it goes. This is a very rich text, just full
of material to keep both student and instructor fascinated.
The exercise sets are rich, having both routine problems (there could be a few
more of these) and advanced problems. The latter are divided into problems
that can be done in an hour or two and those that can lead to protracted projects.
Each is clearly labeled.
If I had to criticize this book for something, I would have to point out that
many of the gures are rather primitive. They were clearly prepared with some
software (such as CorelDraw), but rather ineptly. I think that the publisher
should have caught this aw and corrected it. Apart from the gures, the book
is very professionally prepared and a pleasure to read. It is only black and
whiteno color photographs or plots here. But that is very much in keeping
with the spirit of the book. This is no-nonsense stuff, and is not meant for the
faint of heart.
1 They

ODE.

do, however, provide a nice little Appendix that provides the missing material on basic

Reviews

475

The authors are great believers in MatLab. This use of software is highly appropriate for an engineering audienceand good for everyone else as well. It is
misleading, in todays world, to teach students that most differential equations
can be solved by hand. Nothing could be further from the truth. It is certainly
an epiphany, for instance, to know that the differential equations that describe
an airfoil (such as the wing of the Boeing 767) are solved by a sophisticated
process of guessing. Dont believe me?Ask Tony Jameson, the guy who did
it.
In sum, this is a traditional math book written in the traditional way. But
it is forward-looking, exciting, and innovative for the American curriculum. It
is mathematically exacting (although, in the interest of brevity and readability,
it falls short of proving everything), and it is also credible from the point of
view of applications. It is eminently readable, in a very straightforward and
no-nonsense manner. I like this book very much, and would be delighted to
teach from it. What higher praise could there be?
Steven G. Krantz, Mathematics Dept., Washington University in St. Louis, Box 1146,
One Brookings Dr., St. Louis, MO 631304899; sk@math.wustl.edu , http://
myprofile.cos.com/krantzs10 .

Banks, John, Valentina Dragan, and Arthur Jones. 2003. Chaos: A Mathematical
Introduction. New York: Cambridge University Press; 306 pp, $40. ISBN
0521531047.
Ott, Edward. 2002. Chaos in Dynamical Systems. 2nd ed. New York: Cambridge
University Press; 490 pp, $140, $60 (P). ISBN 0521811961, 0521010845.
Much of science is protected from public scrutiny by the recondite nature
of the enterprise. There is so much technical jargon, and the techniques are so
difcult and abstruse, that the layperson has no hope of beginning to penetrate
even the epidermis of a subject like epidemiology or high-energy physics. But
there are occasional chinks in the armor. And sometimes these occur because,
just by happenstance, a serious scientic subject is given a name that sounds
like something that would interest the butcher or the baker. Examples of this
phenomenon are relativity theory, the uncertainty principle, and chaos.
Flappers and playboys in the Roaring Twenties carried on witty cocktailparty badinage about the social relevance of relativity theory. Everything is
relative but Im sure glad that Im not related to you became an over-used
schoolyard joke. No less an eminence grise than Carl Gustav Jung became obsessed with the lack of verisimilitude that the Heisenberg uncertainty principle
suggested; he develops his ideas in extenso in Jung [1955]. He tells in particular
of his memorable sh day. The story is that he got up that morning and his
wife served him kippers for breakfast. Walking to work, he tripped and fell
over a cod that the shmonger had dropped on the sidewalk. Arriving at the
ofce, he met an old patient who bore a gift for hima wall hanging depicting

476

The UMAP Journal 24.4 (2003)

a sh. And on it went. Jung had a total of seven sh incidents that day. His
conclusion was that this was a manifestation of the uncertainty principlefor
what were the odds that such a conuence of seafood would actually happen?2
Today we enjoy similar exhilaration over chaos. The movie Jurassic Park
features a chaoticist (played by Jeff Goldblum). The uneducated are fascinated by the possibility that the ap of a buttery wing in Brazil could cause a
tsunami in Japannever mind that the effects of the wing ap would dissipate
in about 12 inches. The notion that life is chaotic, and that the physicists can
prove it to be so, evokes a certain giddy delirium.
About 15 years ago, the fractal theorists were fond of saying that fractals are
the footprint of chaos. Assuming that chaos theory has some legitimacy, such
a claim would lend credibility to fractal theory. However, even a cursory look
at phase space shows the allegation to be hogwash. Nevertheless, the idea that
very small perturbations in initial conditions can cause wild deviations in longterm behavior of solutions to differential equations (this is the working mans
denition of chaos) is both physically important and captivating. Surely this
concept is worthy of study.
From my own solipsistic point of view, the chaos/fractal phenomenon is
a remarkable chapter in our social history. As we know, the world is full of
mathematical amateurs. And Im not talking about latent Fermats. Rather, I
speak of people with some mathematical talent but not a lot of training. They
would like to think that they could keep up with a bit of modern mathematics, but there really is no hope. There are not many of useven professional
mathematicianswho can get past the rst page of Andrew Wiless proof of
Fermats Last Theorem, perhaps fewer who will be able to comprehend Grisha
Perelmans putative proof of the Poincare conjecture. Imagine the frustration
of the mathematical amateur. But fractals and chaos are special. They are
obviously something new and hotIBM used to feature them on its television adsand something that practically anyone can understand. Better still,
almost anyone can boot up software and play around with fractals and begin to convince themselves that they are doing mathematics. No wonder that
Heinz Otto Peitgen and Benoit Mandelbrot seem so much like the Pied Piper
of Hamelin: they are chasing us Ph.D. rats out of town and establishing power
for the mathematical people.
In any event, chaos is a facade for a serious set of ideas, and that is the theory
of dynamical systems. We can be grateful that the book of Banks, Dragan, and
Jones (BDJ) presents the subject in just that way: Their title may be Chaos, but
the book is about iteration of functions, bifurcations, periodic orbits, sensitive
dependence on initial data, and all the usual ideas that we associate with this
solid and central mathematical enterprise.
The book BDJ has several notable features. The authors wish to present their
subject to students with minimal preparationeven students who dont quite
know what a proof is. The only prerequisites for reading this book are calculus
2 Persi

Diaconis has set up a probabilistic model for this situation. He has calculated that in a
lifetime of about 40 years, the probability of a seven-fold coincidence like this is about .5.

Reviews

477

and a dollop of mathematical sophistication. The authors profess that one of


the joys of teaching this material to an audience of this sort is that the subject
is current, exciting, and has manifold applications. In point of fact, BDJ says
little about applications (contrast, for example, the lovely book [STR]), and also
gives little sense of the history of, or the immediacy of, the ideas that they are
presenting. Most of their historical allusions are paeans to Poincarewhich is
certainly ne with me. I simply think that BDJ perhaps promised to deliver a
bit more than they were actually able to muster.
The book BDJ has much to recommend it:
crisp, clean graphics without a lot of distracting and irrelevant pictures of
fractals;
lots of accessible and carefully-worked-out examples; and
lucid, well-organized, and well-written prose.
Of utmost importanceand something that I admire greatly in good textbook
writersis that the authors have a clear sense of their audience and are consistent in addressing that audience. There are some mathematical novelties
compared to other texts in the subject. For example, the authors show how
to use the Schwarzian derivative (a moderately abstruse idea from complex
function theory) to detect chaos. Most of the material in BDJ is down-to-earth
and straightforward and just right for the untutored audience that the authors
have in mind. The book is well-organized and well-planned and would be a
pleasure to use in the classroom.
The book by Ott is a different sort of effort. First of all, it is the second edition
of a classic text. It is considerable more ambitious than BDJ, having over 50%
more pages. It covers many more classical and well-established (some might
say more central) topics. These include
the Smale horseshoe,
Lyapunov exponents,
stable and unstable manifolds,
higher-dimensional systems,
chaotic scattering, and
control and synchronization of chaos.
There are many more. Although Ott fails to describe the intended audience,
one deduces that this book is meant for a rather more sophisticated audience
than the readers of BDJ. Otts book has fewer examples, fewer elementary
calculations, and more rigorous and thorough proofs and discussions than
those that can be found in BDJ. The prerequisites for Otts book (which, more
is the pity, he also does not bother to enunciate for us) appear to include linear

478

The UMAP Journal 24.4 (2003)

algebra, perhaps some real analysis, and also some knowledge of differential
equations.
Otts book shares some of the nice features of BDJ: clean and effective graphics, good exercise sets, and clear writing. Ott has end-of-chapter notes, a nice
scholarly feature that is both enlightening and entertaining. He has a much
more thorough bibliography than does BDJ.
Both BDJ and Ott are serious efforts by serious scholars to bring students
into the fold of chaos study. I would say that Ott brings his readers much closer
to the re. BDJ reach down to a pretty naive audience and inculcate them with
some mathematical literacy. Ott takes that literacy for granted and pushes his
followers quite a bit further.
Both these books are worthy implementations of the text-writing art. Both
would be rewarding to use in a class. Both are reliable and thorough and timely.
I can recommend them with pleasure.

References
Jung, C.G. 1955. Synchronicity: An Acausal Connectivity Principle. New York:
Pantheon Books.
Strogatz, S.H. 1994. Nonlinear Dynamics and Chaos. Reading, MA: AddisonWesley.
Steven G. Krantz, Mathematics Dept., Washington University in St. Louis, Box 1146,
One Brookings Dr., St. Louis, MO 631304899; sk@math.wustl.edu , http://
myprofile.cos.com/krantzs10 .

Kaplan, Wilfred. 1999. Maxima and Minima with Applications: Practical Optimization and Duality. New York: Wiley. x + 284, $99.95. ISBN 0471252891.
We have been lucky to get Prof. Kaplan (b. 1915) to review a couple of books
recently. Unfortunately, the eminent scholar who agreed to review this book
by Kaplan has not reported; but lacking a review, I offer a brief mention here,
which the book denitely merits. While it is accessible to undergraduates, it is
ideal for many graduate students and should be regarded as a graduate text.
Any graduate student of operations research will almost certainly benet from
the book, as will anyone with a serious interest in optimization.
James M. Cargal, Mathematics Dept., Troy State University Montgomery, Montgomery, AL 361210667; jmcargal@sprintmail.com .

Annual Index

479

Annual Index for Vol. 24, 2003


Author Index
Acknowledgments. 24(4): 483.
Annual Index. 24(4): 479482.
Arney, Chris. Results of the 2003 Interdisciplinary Contest in Modeling. 24(1): 97110.
Ashline, George, Alain Brizard, and Joanna Ellis-Monaghan. Water rockets in ight:
Calculus in action. UMAP/ILAP Modules 200203: Tools for Teaching: 151188.
Banister, Melissa J., Matthew Macauley, and Micah J. Smukler. Thinking outside the
box and over the elephant. 24(3): 253264.
Berkove, Ethan, Thomas Hill, and Scott Moor. Getting the salt out. 24(4): 417431.
Blunk, Mark. See Winstrom, Luke.
Brizard, Alain Brizard. See Ashline, George.
Campbell, Paul J. Acknowledgments. 24(4): 483.
. Environmental modeling: Not just a senior elective any more. 24(2): 9395.
. Explore the universe: Through issues of The UMAP Journal. 24(1): 12.
. Guide for Authors. 24(1): 9192; UMAP/ILAP Modules 200203: Tools for
Teaching: 189190.
. We have already been to the future. 24(4): 393394.
. See Garfunkel, Sol.
Cassady, C. Richard. Judges Commentary: The Outstanding Airport Screening papers.
24(2): 185188.
Corwin, Ivan, Sheel Ganatra, and Nikita Rozenblyum. A time-independent model of
box safety for stunt motorcyclists. 24(3): 235252.
Coskey, Sam. See Winstrom, Luke.
Deng Xiaowei, Xu Wei, and Zhang Zhenyu. You too can be James Bond. 24(3): 265282.
Diaz, Carlos A. See Livesey, Michelle R.
Edlund, Jeffrey A. See Kishimoto, Chad T.
Ellis-Monaghan, Joanna. See Ashline, George.
Errata. 24(4): 484.
Esser, Ernie. See Giansiracusa, Jeffrey.
Fitzkee, Thomas L. Tuition prepayment plan. 24(1): 5378.
Fox, William P. Judges Commentary: The Outstanding Stunt Person papers. 24(3):
319324.
. See Pike, Lisa.
Ganatra, Sheel Ganatra. See Corwin, Ivan.
Garfunkel, Sol, and Paul J. Campbell. Introduction. UMAP/ILAP Modules 200203: Tools
for Teaching: vvi.
. Secondaryundergraduate articulation: Moving forward 24(3): 193198.
Giansiracusa, Jeffrey, Ernie Esser, and Simon Pai. Cardboard comfortable when it comes
to crashing. 24(3): 283300.

480

The UMAP Journal 24.4 (2003)

Gillis, Darin W., David R. Lindstone, and Aaron T. Windeld. The gamma knife problem.
24(3): 353365.
Giordano, Frank. Results of the 2003 Contest in Mathematical Modeling. 24(3): 199220.
Greenberg, Kyle Andrew. See Powell, Michael Alan.
Grove, Sarah, Chris Jones, and Joel Lepak. Shoot to kill! 24(3): 381390.
Guide for Authors. 24(1): 9192; 24(2): 195196. UMAP/ILAP Modules 200203: Tools
for Teaching: 189190.
Haining, Robert T., Dana M. Lindemann, and Neal P. Richardson. Feds with EDS:
Searching for the optimal explosive scanning system. 24(2): 169184.
Hill, Thomas. See Berkove, Ethan.
Hu Yuxiao, Hua Zheng, and Zhou Enlu. Fly with Condence. 24(3): 301318.
Hua Zheng. See Hu Yuxiao.
Hughes, Frank. See Scharf, John L.
Isihara, Paul Atsusi, Benjamin Noonan, and Nathaniel Stapleton. Liouvilles theorem
in dynamics. UMAP/ILAP Modules 200203: Tools for Teaching: 4968.
Jacobson, Sheldon H., and John E. Kobza. Authors Commentary: Aviation security
baggage screening strategies: To screen or not to screen, that is the question! 24(2):
189194.
Jarrow, Tate Alan. See Powell, Michael Alan.
Johnson, Kylan Neal. See Olson, Gary Allen.
Kao, Justin C. See Kishimoto, Chad T.
Kelly, Susan E., and James S. Walker. Efcient signal transmission and wavelet-based
compression. 24(4): 395416.
Kishimoto, Chad T., Justin C. Kao, and Jeffrey A. Edlund. Safe landings. 24(3): 221234.
Kobza, John E. See Jacobson, Sheldon H.
Lindemann, Dana M. See Haining, Robert T.
Lindstone, David R. See Gillis, Darin W..
Liukkonen, John R., and Laura J. Steinberg. Chloroform alert! University students
exposed to unhealthy air. UMAP/ILAP Modules 200203: Tools for Teaching: 105124.
Liukonnen, John R. [misspelled name] See Liukkonen, John R.
Livesey, Michelle R., Carlos A. Diaz, and Terrence K. Williams. Advancing airport
security through optimization and simulation. 24(2): 141152.
Long Yun, Ye Yungqing, and Wei Zhen. A sphere-packing model for the optimal treatment plan. 24(3): 341352.
Martin, Tara, Gautam Thatte, and Michael Vrable. How I learned to stop worrying and
nd the bomb. 24(2): 123140.
Macauley, Matthew. See Banister, Melissa J.
Mitchell, Kevin, and James Ryan. Game theory models of animal behavior. UMAP/ILAP
Modules 200203: Tools for Teaching: 148.
Moor, Scott. See Berkove, Ethan.
Noonan, Benjamin. See Isihara, Paul Atsusi.

Annual Index

481

Olson, Gary Allen, Kylan Neal Johnson, and Joseph Paul Rasca. Airport baggage screening: Optimizing the implementation of EDS machines. 24(2): 111122.
Pai, Simon Pai. See Giansiracusa, Jeffrey.
Pike, Lisa, and William P. Fox. Stocking a sh pond. UMAP/ILAP Modules 200203:
Tools for Teaching: 7790.
. Survival of early Americans. UMAP/ILAP Modules 200203: Tools for Teaching:
91104.
Powell, Michael Alan, Tate Alan Jarrow, and Kyle Andrew Greenberg. The price of
security: A cost-benet analysis of screening of checked baggage. 24(2): 153168.
Puckette, Emily E. The spread of forest res. 24(4): 417433.
Rasca, Joseph Paul. See Olson, Gary Allen.
Reviews. 24(1): 7990; 24(4): 473478.
Reviews Index. 23(4): 482.
Richardson, Neal P. See Haining, Robert T.
Rozenblyum, Nikita. See Corwin, Ivan.
Ryan, James. See Mitchell, Kevin.
Scharf, John L., and Frank Hughes. Going into orbitLaunching the Shuttle. UMAP/ILAP
Modules 200203: Tools for Teaching: 125150.
Sigmon, Neil P. Determination of satellite orbits with vector calculus. 24(1): 2752.
Small, Don. See Yu, Lei.
Smukler, Micah J. See Banister, Melissa J.
Stapleton, Nathaniel. See Isihara, Paul Atsusi.
Statement of Ownership, Management, and Circulation. 24(3): [391][392].
Steinberg, Laura J. See Liukkonen, John R.
Sun Fei, Yang Lin, and Wang Hong. The genetic algorithm-based optimization approach
for gamma unit treatment. 24(3): 325340.
Teets. Donald A. Predicting opportunities for viewing the International Space Station.
24(1): 326.
Thatte, Gautam. See Martin, Tara.
Vanisko, Marie. Who falls through the healthcare safety net? UMAP/ILAP Modules
200203: Tools for Teaching: 6976.
Vrable, Michael. See Martin, Tara.
Walker, James S. See Kelly, Susan E.
Wang Hong. See Sun Fei.
Wei Zhen.See Long Yun.
Williams, Terrence K. See Livesey, Michelle R.
Windeld, Aaron T. See Gillis, Darin W.
Winstrom, Luke, Sam Coskey, and Mark Blunk. Shelling tumors with caution and
wiggles. 24(3): 367380.
Xu Wei. See Deng Xiaowei.
Yang Lin. See Sun Fei.

482

The UMAP Journal 24.4 (2003)

Ye Yungqing. See Long Yun.


Yu, Lei, and Don Small. Vehicle emissions. 24(4): 451472.
Zhang Zhenyu. See Deng Xiaowei.
Zhou Enlu. See Hu Yuxiao.

Reviews Index
(Names of authors are in plain type; names of reviewers are in bold.)
Banks, John, Valentina Dragan, and Arthur Jones. Chaos: A Mathematical Introduction.
Steven G. Krantz. 24(4): 475478.
Biggs, Norman L. 1995. Discrete Mathematics. 2nd ed. James M. Cargal. 24(1): 8487.
Giordano, Frank R., Maurice D. Weir, and William P. Fox. A First Course in Mathematical
Modeling. 3rd ed. David Bressoud. 24(1): 7980.
Gordon, Hugh. Discrete Probability. James M. Cargal. 24(1): 8487.
Haggarty, Rod. Discrete Mathematics for Computing. James M. Cargal. 24(1): 8487.
Hubbard, John H., and Barbara Burke Hubbard. Vector Calculus, Linear Algebra, and
Differential Forms: A Unied Approach. 2nd ed. David Bressoud. 24(1): 81.
James, J.F. 1995. A Students Guide to Fourier Transforms: With Applications in Physics and
Engineering. Steven G. Krantz. 24(1): 8283.
Kaplan, Wilfred. Maxima and Minima with Applications: Practical Optimization and Duality.
James M. Cargal. 24(4): 478.
King, A.C., J. Billingham, and S.R. Otto. Differential Equations: Linear, Nonlinear, Ordinary,
Partial. Steven G. Krantz. 24(4): 473475.
Martin, George E. The Art of Enumerative Combinatorics. James M. Cargal. 24(1): 8487.
Maurer, Stephen B., and Anthony Ralston. Discrete Algorithmic Mathematics. 2nd ed.
James M. Cargal. 24(1): 8487.
Ott, Edward. Chaos in Dynamical Systems. 2nd ed. Steven G. Krantz. 24(4): 475478.
Ross, Sheldon M. Topics in Finite and Discrete Mathematics. James M. Cargal. 24(1):
8487.
Salsburg, David. The Lady Tasting Tea. James M. Cargal. 24(1): 8790.
Sandefur, James T. 2003. Elementary Mathematical Modeling: A Dynamic Approach. David
Bressoud. 24(1): 7980.
Tucker, Alan. 1995. Applied Combinatorics. 3rd ed. James M. Cargal. 24(1): 8487.

Annual Index

483

Acknowledgments
I would like to express my great appreciation for the help of the associate editors,
whose names appear on the masthead. Not only do they do the bulk of work of evaluating manuscripts, but they also solicit new works and encourage and guide potential
authors.
I am also indebted to the additional individuals listed below who have reviewed
manuscripts during the past year. Their careful evaluation and judgment have enhanced
the quality of the articles and Modules that have appeared in the Journal and in Tools
for Teaching. (Reviewers of some papers considered for Vol. 24 of the Journal were
acknowledged already in Vol. 23, No. 4. Reviewers and editors of the ILAP Modules in
UMAP/ILAP Modules 200203: Tools for Teaching are acknowledged on the frontispiece
of the corresponding ILAP Module.)
The Journal offers many opportunities for participation in COMAPs work. If you
would like to
referee manuscriptsplease contact me;
review books, software, or lmsplease contact the Reviews Editor;
write a self-contained expository essay about an area of mathematics, however grand
or smallan era, a concept, a theorem, an idea, a termplease contact the On Jargon
Editor;
contribute an Interdisciplinary Lively Applications Project (ILAP) Moduleplease
contact the ILAP Editor;
contribute an article, UMAP Module, or Minimoduleplease contact me;
encourage and stimulate colleagues to prepare and submit suitable materialplease
contact me about being nominated to join the Editorial Board.
Contact information for the associate editors in charge of Reviews, On Jargon, and
ILAPs, as well as my own information, are on the masthead of every issue.
Finally, the associate editors and I would like to thank the Journals authors, without
whom none of this would be possible, and its readers, whose benet and enjoyment are
the culmination of our enterprise.
Paul J. Campbell, Editor
Bruce Atwood, Beloit College
Eric Bach, University of Wisconsin
Steven J. Brams, New York University
Russell Goodman and Christy Nichting, Central College
Steven Janke, Colorado College
Herve Moulin, Rice University
Fred Solomon, Warren Wilson College
Paul Stanley, Beloit College
Alan Taylor, Union College (NY)
David Vessey, Beloit College
Jack Winn, SUNY at Farmingdale
David Wolfe, Gustavus Adolphus College

484

The UMAP Journal 24.4 (2003)

Errata
Vol. 23, No. 1 (2002)
p. 76, , l. 18: group The group
Vol. 23, No. 2 (2002)
p. 124, l. 1: p

n1
i=0


m1 k 1 + i i
k1+i i
k
q p
q
i=0
i
i

p. 126, l. 10: 1/3 1/4


p. 131, ll. 3, 6, and 8: delete last two ]] in each assignment
p. 132, l. 18: (If F>0, (If [F>0,
p. 133, l. 10: should read
Basket[a_] := If[a[[1]]>0,
SortInOrder
[If[SortInOrder[Sack[a]][[1]]>0,
SortInOrder[Sack[SortInOrder[Sack[a]]]]
Sack[a]]],
a]
or alternatively and more simply via recursion,
Basket[a_} := Decreaser[[Decreaser[a]]
where
Decreaser[a_]:=If[a[[1]]>0,SortInOrder[Sack[a]],a]
p. 148, l. 3, and p. 160, l. 11: Franics Francis
p. 163, sidebar, and p. 182, l. 9: Liukonnen Liukkonen
p. 172, l. 12, eq. (16): , dr dr,
Vol. 23, No. 4 (2002)
p. 416: The following reference was omitted:
Thayer, William T. 1983. The St. Louis Arch. UMAP Modules in Undergraduate
Mathematics and Its Applications: Module 638. The UMAP Journal 4(3): 347
365. 1984. Reprinted in UMAP Modules 1983: Tools for Teaching, 446464.
Lexington, MA: COMAP.
p. 456, l. 2: insert afliaton: Kansas State University
UMAP/ILAP Modules 20022003: Tools for Teaching
p. [iii] [Table of Contents], l. 7; p. 124, l. 9; and back cover, l. 7:
Liukonnen Liukkonen
Vol. 24, No. 1 (2003)
p. 54, l. 10: delete: (see Table 1)
p. 79, l. 3: delete: Elementary
Vol. 24, No. 3 (2003)
p. 193: Pagination of this issue should have begun with p. 197.
p. 265, Abstract, l. 14: that assume assume that
p. 271, l. 12: 2/3 1/3.
p. 274, l. 1: P0 (V0 0 dt P0 (V0 0 dt)
p. 277, l. 3: hHpile h Hpile